refactor(auth): simplify OSS authorization to bare minimum

OSS module (openhands/server/auth/):
- Permission enum only
- Single no-op require_permission() that returns user_id
- No roles, no ROLE_PERMISSIONS, no helper methods

Enterprise module (enterprise/server/auth/authorization.py):
- Imports Permission from OSS module
- Defines RoleName enum and ROLE_PERMISSIONS locally
- Implements all authorization logic: require_permission(),
  require_org_role(), has_permission(), get_role_permissions(), etc.

Co-authored-by: openhands <openhands@all-hands.dev>

.github/ISSUE_TEMPLATE/bug_template.yml

@@ -5,52 +5,113 @@ labels: ['bug']
 body:
   - type: markdown
     attributes:
-      value: Thank you for taking the time to fill out this bug report. Please provide as much information as possible
-       to help us understand and address the issue effectively.
+      value: |
+        ## Thank you for reporting a bug! 🐛
+
+        **Please fill out all required fields.** Issues missing critical information (version, installation method, reproduction steps, etc.) will be delayed or closed until complete details are provided.
+
+        Clear, detailed reports help us resolve issues faster.
 
   - type: checkboxes
     attributes:
-      label: Is there an existing issue for the same bug? (If one exists, thumbs up or comment on the issue instead).
-      description: Please check if an issue already exists for the bug you encountered.
+      label: Is there an existing issue for the same bug?
+      description: Please search existing issues before creating a new one. If found, react or comment to the duplicate issue instead of making a new one.
       options:
-      - label: I have checked the existing issues.
+      - label: I have searched existing issues and this is not a duplicate.
         required: true
 
   - type: textarea
     id: bug-description
     attributes:
-      label: Describe the bug and reproduction steps
-      description: Provide a description of the issue along with any reproduction steps.
+      label: Bug Description
+      description: Clearly describe what went wrong. Be specific and concise.
+      placeholder: Example - "When I run a Python task, OpenHands crashes after 30 seconds with a connection timeout error."
     validations:
       required: true
 
+  - type: textarea
+    id: expected-behavior
+    attributes:
+      label: Expected Behavior
+      description: What did you expect to happen?
+      placeholder: Example - "OpenHands should execute the Python script and return results."
+    validations:
+      required: false
+
+  - type: textarea
+    id: actual-behavior
+    attributes:
+      label: Actual Behavior
+      description: What actually happened?
+      placeholder: Example - "Connection timed out after 30 seconds, task failed with error code 500."
+    validations:
+      required: false
+
+  - type: textarea
+    id: reproduction-steps
+    attributes:
+      label: Steps to Reproduce
+      description: Provide clear, step-by-step instructions to reproduce the bug.
+      placeholder: |
+        1. Install OpenHands using Docker
+        2. Configure with Claude 3.5 Sonnet
+        3. Run command: `openhands run "write a python script"`
+        4. Wait 30 seconds
+        5. Error appears
+    validations:
+      required: false
+
   - type: dropdown
     id: installation
     attributes:
-      label: OpenHands Installation
+      label: OpenHands Installation Method
       description: How are you running OpenHands?
       options:
-        - Docker command in README
-        - GitHub resolver
+        - CLI (uv tool install)
+        - CLI (executable binary)
+        - CLI (Docker)
+        - Local GUI (Docker web interface)
+        - OpenHands Cloud (app.all-hands.dev)
+        - SDK (Python library)
         - Development workflow
-        - CLI
-        - app.all-hands.dev
         - Other
       default: 0
+    validations:
+      required: false
+
+  - type: input
+    id: installation-other
+    attributes:
+      label: If you selected "Other", please specify
+      description: Describe your installation method
+      placeholder: ex. Custom Kubernetes deployment, pip install from source, etc.
 
   - type: input
     id: openhands-version
     attributes:
       label: OpenHands Version
-      description: What version of OpenHands are you using?
-      placeholder: ex. 0.9.8, main, etc.
+      description: What version are you using? Find this in settings or by running `openhands --version`
+      placeholder: ex. 0.9.8, main, commit hash, etc.
+    validations:
+      required: false
+
+  - type: checkboxes
+    id: version-confirmation
+    attributes:
+      label: Version Confirmation
+      description: Bugs on older versions may already be fixed. Please upgrade before submitting.
+      options:
+        - label: "I have confirmed this bug exists on the LATEST version of OpenHands"
+          required: false
 
   - type: input
     id: model-name
     attributes:
       label: Model Name
-      description: What model are you using?
-      placeholder: ex. gpt-4o, claude-3-5-sonnet, openrouter/deepseek-r1, etc.
+      description: Which LLM model are you using?
+      placeholder: ex. gpt-4o, claude-3-5-sonnet-20241022, openrouter/deepseek-r1, etc.
+    validations:
+      required: false
 
   - type: dropdown
     id: os
@@ -60,12 +121,46 @@ body:
         - MacOS
         - Linux
         - WSL on Windows
+        - Windows (Docker Desktop)
+        - Other
+    validations:
+      required: false
+
+  - type: input
+    id: browser
+    attributes:
+      label: Browser (if using web UI)
+      description: |
+        If applicable, which browser and version?
+
+      placeholder: ex. Chrome 131, Firefox 133, Safari 17.2
+
+  - type: textarea
+    id: logs
+    attributes:
+      label: Logs and Error Messages
+      description: |
+        **Paste relevant logs, error messages, or stack traces.** Use code blocks (```) for formatting.
+
+        LLM logs are in `logs/llm/default/`. Include timestamps if errors occurred at a specific time.
+      placeholder: |
+        ```
+        Paste error logs here
+        ```
 
   - type: textarea
     id: additional-context
     attributes:
-      label: Logs, Errors, Screenshots, and Additional Context
-      description: Please provide any additional information you think might help. If you want to share the chat history
-        you can click the thumbs-down (👎) button above the input field and you will get a shareable link
-        (you can also click thumbs up when things are going well of course!). LLM logs will be stored in the
-        `logs/llm/default` folder. Please add any additional context about the problem here.
+      label: Screenshots and Additional Context
+      description: |
+        Add screenshots, videos, runtime environment, or other context that helps explain the issue.
+
+        💡 **Share conversation history:** In the OpenHands chat UI, click the 👎 or 👍 button (above the message input) to generate a shareable link to your conversation.
+
+      placeholder: Drag and drop screenshots here, paste links, or add additional context.
+
+  - type: markdown
+    attributes:
+      value: |
+        ---
+        **Note:** Issues with incomplete information may be closed or deprioritized. Maintainers and community members have limited bandwidth and prioritize well-documented bugs that are easier to reproduce and fix. Thank you for your understanding!

.github/ISSUE_TEMPLATE/feature_request.md

@@ -1,17 +0,0 @@
----
-name: Feature Request or Enhancement
-about: Suggest an idea for an OpenHands feature or enhancement
-title: ''
-labels: 'enhancement'
-assignees: ''
-
----
-
-**What problem or use case are you trying to solve?**
-
-**Describe the UX or technical implementation you have in mind**
-
-**Additional context**
-
-
-### If you find this feature request or enhancement useful, make sure to add a 👍 to the issue

.github/ISSUE_TEMPLATE/feature_request.yml

@@ -0,0 +1,105 @@
+name: Feature Request or Enhancement
+description: Suggest a new feature or improvement for OpenHands
+title: '[Feature]: '
+labels: ['enhancement']
+body:
+  - type: markdown
+    attributes:
+      value: |
+        ## Thank you for suggesting a feature! 💡
+
+        **Please provide detailed information.** Vague or low-effort requests may be closed. Well-documented feature requests with strong community support are more likely to be added to the roadmap.
+
+  - type: checkboxes
+    attributes:
+      label: Is there an existing feature request for this?
+      description: Please search existing issues and feature requests before creating a new one. If found, react or comment to the duplicate issue instead of making a new one.
+      options:
+      - label: I have searched existing issues and feature requests, and this is not a duplicate.
+        required: true
+
+  - type: textarea
+    id: problem-statement
+    attributes:
+      label: Problem or Use Case
+      description: What problem are you trying to solve? What use case would this feature enable?
+      placeholder: |
+        Example - "As a developer working on large codebases, I need to search across multiple files simultaneously. Currently, I have to search file-by-file which is time-consuming and inefficient."
+    validations:
+      required: true
+
+  - type: textarea
+    id: proposed-solution
+    attributes:
+      label: Proposed Solution
+      description: Describe your ideal solution. What should this feature do? How should it work?
+      placeholder: |
+        Example - "Add a global search feature that allows searching across all files in the workspace. Results should show file name, line number, and context around matches. Include regex support and filtering options."
+    validations:
+      required: true
+
+  - type: textarea
+    id: alternatives
+    attributes:
+      label: Alternatives Considered
+      description: Have you considered any alternative solutions or workarounds? What are their limitations?
+      placeholder: Example - "I tried using grep in the terminal, but it's not integrated with the UI and doesn't provide click-to-navigate functionality."
+
+  - type: dropdown
+    id: priority
+    attributes:
+      label: Priority / Severity
+      description: How important is this feature to your workflow?
+      options:
+        - "Critical - Blocking my work, no workaround available"
+        - "High - Significant impact on productivity"
+        - "Medium - Would improve experience"
+        - "Low - Nice to have"
+      default: 2
+    validations:
+      required: true
+
+  - type: dropdown
+    id: scope
+    attributes:
+      label: Estimated Scope
+      description: To the best of your knowledge, how complex do you think this feature would be to implement?
+      options:
+        - "Small - UI tweak, config option, or minor change"
+        - "Medium - New feature with moderate complexity"
+        - "Large - Significant feature requiring architecture changes"
+        - "Unknown - Not sure about the technical complexity"
+      default: 3
+
+  - type: dropdown
+    id: feature-area
+    attributes:
+      label: Feature Area
+      description: Which part of OpenHands does this feature relate to? If you select "Other", please specify the area in the Additional Context section below.
+      options:
+        - "Agent / AI behavior"
+        - "User Interface / UX"
+        - "CLI / Command-line interface"
+        - "File system / Workspace management"
+        - "Configuration / Settings"
+        - "Integrations (GitHub, GitLab, etc.)"
+        - "Performance / Optimization"
+        - "Documentation"
+        - "Other"
+    validations:
+      required: true
+
+  - type: textarea
+    id: technical-details
+    attributes:
+      label: Technical Implementation Ideas (Optional)
+      description: If you have technical expertise, share implementation ideas, API suggestions, or relevant technical details.
+      placeholder: |
+        Example - "Could use ripgrep library for fast search. Expose results via /api/search endpoint. Frontend can use virtualized list for rendering large result sets."
+
+  - type: textarea
+    id: additional-context
+    attributes:
+      label: Additional Context
+      description: Add any other context, screenshots, mockups, or examples that help illustrate this feature request.
+      placeholder: Drag and drop screenshots, mockups, or links here.

.github/pull_request_template.md

@@ -1,4 +1,4 @@
-<!-- Ideally you should open a PR when it is ready for review. Draft PRs will not be reviewed -->
+<!-- If you are still working on the PR, please mark it as draft. Maintainers will review PRs marked ready for review, which leads to lost time if your PR is actually not ready yet. Keep the PR marked as draft until it is finally ready for review -->
 
 ## Summary of PR
 

.github/workflows/ghcr-build.yml

@@ -9,6 +9,7 @@ on:
   push:
     branches:
       - main
+      - "saas-rel-*"
     tags:
       - "*"
   pull_request:

.github/workflows/pr-review-by-openhands.yml

@@ -0,0 +1,127 @@
+---
+name: PR Review by OpenHands
+
+on:
+    # Use pull_request_target to allow fork PRs to access secrets when triggered by maintainers
+    # Security: This workflow runs when:
+    #   1. A new PR is opened (non-draft), OR
+    #   2. A draft PR is marked as ready for review, OR
+    #   3. A maintainer adds the 'review-this' label, OR
+    #   4. A maintainer requests openhands-agent or all-hands-bot as a reviewer
+    # Only users with write access can add labels or request reviews, ensuring security.
+    # The PR code is explicitly checked out for review, but secrets are only accessible
+    # because the workflow runs in the base repository context
+    pull_request_target:
+        types: [opened, ready_for_review, labeled, review_requested]
+
+permissions:
+    contents: read
+    pull-requests: write
+    issues: write
+
+jobs:
+    pr-review:
+        # Run when one of the following conditions is met:
+        #   1. A new non-draft PR is opened by a trusted contributor, OR
+        #   2. A draft PR is converted to ready for review by a trusted contributor, OR
+        #   3. 'review-this' label is added, OR
+        #   4. openhands-agent or all-hands-bot is requested as a reviewer
+        # Note: FIRST_TIME_CONTRIBUTOR PRs require manual trigger via label/reviewer request
+        if: |
+            (github.event.action == 'opened' && github.event.pull_request.draft == false && github.event.pull_request.author_association != 'FIRST_TIME_CONTRIBUTOR') ||
+            (github.event.action == 'ready_for_review' && github.event.pull_request.author_association != 'FIRST_TIME_CONTRIBUTOR') ||
+            github.event.label.name == 'review-this' ||
+            github.event.requested_reviewer.login == 'openhands-agent' ||
+            github.event.requested_reviewer.login == 'all-hands-bot'
+        concurrency:
+            group: pr-review-${{ github.event.pull_request.number }}
+            cancel-in-progress: true
+        runs-on: blacksmith-4vcpu-ubuntu-2404
+        env:
+            LLM_MODEL: litellm_proxy/claude-sonnet-4-5-20250929
+            LLM_BASE_URL: https://llm-proxy.app.all-hands.dev
+            # PR context will be automatically provided by the agent script
+            PR_NUMBER: ${{ github.event.pull_request.number }}
+            PR_TITLE: ${{ github.event.pull_request.title }}
+            PR_BODY: ${{ github.event.pull_request.body }}
+            PR_BASE_BRANCH: ${{ github.event.pull_request.base.ref }}
+            PR_HEAD_BRANCH: ${{ github.event.pull_request.head.ref }}
+            REPO_NAME: ${{ github.repository }}
+        steps:
+            - name: Checkout software-agent-sdk repository
+              uses: actions/checkout@v5
+              with:
+                  repository: OpenHands/software-agent-sdk
+                  path: software-agent-sdk
+
+            - name: Checkout PR repository
+              uses: actions/checkout@v5
+              with:
+                  # When using pull_request_target, explicitly checkout the PR branch
+                  # This ensures we review the actual PR code (including fork PRs)
+                  repository: ${{ github.event.pull_request.head.repo.full_name }}
+                  ref: ${{ github.event.pull_request.head.ref }}
+                  fetch-depth: 0
+                  # Security: Don't persist credentials to prevent untrusted PR code from using them
+                  persist-credentials: false
+                  path: pr-repo
+
+            - name: Set up Python
+              uses: actions/setup-python@v6
+              with:
+                  python-version: '3.13'
+
+            - name: Install uv
+              uses: astral-sh/setup-uv@v7
+              with:
+                  enable-cache: true
+
+            - name: Install GitHub CLI
+              run: |
+                  # Install GitHub CLI for posting review comments
+                  sudo apt-get update
+                  sudo apt-get install -y gh
+
+            - name: Install OpenHands dependencies
+              run: |
+                  # Install OpenHands SDK and tools from local checkout
+                  uv pip install --system ./software-agent-sdk/openhands-sdk ./software-agent-sdk/openhands-tools
+
+            - name: Check required configuration
+              env:
+                  LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
+              run: |
+                  if [ -z "$LLM_API_KEY" ]; then
+                    echo "Error: LLM_API_KEY secret is not set."
+                    exit 1
+                  fi
+
+                  echo "PR Number: $PR_NUMBER"
+                  echo "PR Title: $PR_TITLE"
+                  echo "Repository: $REPO_NAME"
+                  echo "LLM model: $LLM_MODEL"
+                  if [ -n "$LLM_BASE_URL" ]; then
+                    echo "LLM base URL: $LLM_BASE_URL"
+                  fi
+
+            - name: Run PR review
+              env:
+                  LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
+                  GITHUB_TOKEN: ${{ secrets.ALLHANDS_BOT_GITHUB_PAT }}
+                  LMNR_PROJECT_API_KEY: ${{ secrets.LMNR_SKILLS_API_KEY }}
+              run: |
+                  # Change to the PR repository directory so agent can analyze the code
+                  cd pr-repo
+
+                  # Run the PR review script from the software-agent-sdk checkout
+                  uv run python ../software-agent-sdk/examples/03_github_workflows/02_pr_review/agent_script.py
+
+            - name: Upload logs as artifact
+              uses: actions/upload-artifact@v5
+              if: always()
+              with:
+                  name: openhands-pr-review-logs
+                  path: |
+                      *.log
+                      output/
+                  retention-days: 7

CONTRIBUTING.md

@@ -6,11 +6,12 @@ Thanks for your interest in contributing to OpenHands! We welcome and appreciate
 
 To understand the codebase, please refer to the README in each module:
 - [frontend](./frontend/README.md)
-- [evaluation](./evaluation/README.md)
 - [openhands](./openhands/README.md)
    - [agenthub](./openhands/agenthub/README.md)
    - [server](./openhands/server/README.md)
 
+For benchmarks and evaluation, see the [OpenHands/benchmarks](https://github.com/OpenHands/benchmarks) repository.
+
 ## Setting up Your Development Environment
 
 We have a separate doc [Development.md](https://github.com/OpenHands/OpenHands/blob/main/Development.md) that tells

Development.md

@@ -200,7 +200,7 @@ Here's a guide to the important documentation files in the repository:
 - [/frontend/README.md](./frontend/README.md): Frontend React application setup and development guide
 - [/containers/README.md](./containers/README.md): Information about Docker containers and deployment
 - [/tests/unit/README.md](./tests/unit/README.md): Guide to writing and running unit tests
-- [/evaluation/README.md](./evaluation/README.md): Documentation for the evaluation framework and benchmarks
+- [OpenHands/benchmarks](https://github.com/OpenHands/benchmarks): Documentation for the evaluation framework and benchmarks
 - [/skills/README.md](./skills/README.md): Information about the skills architecture and implementation
 - [/openhands/server/README.md](./openhands/server/README.md): Server implementation details and API documentation
 - [/openhands/runtime/README.md](./openhands/runtime/README.md): Documentation for the runtime environment and execution model

config.template.toml

@@ -440,12 +440,6 @@ type = "noop"
 #temperature = 0.1
 #max_input_tokens = 1024
 
-#################################### Eval ####################################
-# Configuration for the evaluation, please refer to the specific evaluation
-# plugin for the available options
-##############################################################################
-
-
 ########################### Kubernetes #######################################
 # Kubernetes configuration when using the Kubernetes runtime
 ##############################################################################

docker-compose.yml

@@ -8,7 +8,7 @@ services:
     container_name: openhands-app-${DATE:-}
     environment:
       - AGENT_SERVER_IMAGE_REPOSITORY=${AGENT_SERVER_IMAGE_REPOSITORY:-ghcr.io/openhands/agent-server}
-      - AGENT_SERVER_IMAGE_TAG=${AGENT_SERVER_IMAGE_TAG:-0fdea73-python}
+      - AGENT_SERVER_IMAGE_TAG=${AGENT_SERVER_IMAGE_TAG:-31536c8-python}
       #- SANDBOX_USER_ID=${SANDBOX_USER_ID:-1234} # enable this only if you want a specific non-root sandbox user but you will have to manually adjust permissions of ~/.openhands for this user
       - WORKSPACE_MOUNT_PATH=${WORKSPACE_BASE:-$PWD/workspace}
     ports:

enterprise/Dockerfile

@@ -23,12 +23,23 @@ RUN apt-get update && \
     apt-get clean && \
     rm -rf /var/lib/apt/lists/*
 
-# Install Python packages with security fixes
-RUN /app/.venv/bin/pip install alembic psycopg2-binary cloud-sql-python-connector pg8000 gspread stripe python-keycloak asyncpg sqlalchemy[asyncio] resend tenacity slack-sdk ddtrace "posthog>=6.0.0" "limits==5.2.0" coredis prometheus-client shap scikit-learn pandas numpy google-cloud-recaptcha-enterprise && \
-    # Update packages with known CVE fixes
-    /app/.venv/bin/pip install --upgrade \
-        "mcp>=1.10.0" \
-        "pillow>=11.3.0"
+# Install poetry and export before importing current code.
+RUN /app/.venv/bin/pip install poetry poetry-plugin-export
+
+# Install Python dependencies from poetry.lock for reproducible builds
+# Copy lock files first for better Docker layer caching
+COPY --chown=openhands:openhands enterprise/pyproject.toml enterprise/poetry.lock /tmp/enterprise/
+RUN cd /tmp/enterprise && \
+    # Export only main dependencies with hashes for supply chain security
+    /app/.venv/bin/poetry export --only main -o requirements.txt && \
+    # Remove the local path dependency (openhands-ai is already in base image)
+    sed -i '/^-e /d; /openhands-ai/d' requirements.txt && \
+    # Install pinned dependencies from lock file
+    /app/.venv/bin/pip install -r requirements.txt && \
+    # Cleanup - return to /app before removing /tmp/enterprise
+    cd /app && \
+    rm -rf /tmp/enterprise && \
+    /app/.venv/bin/pip uninstall -y poetry poetry-plugin-export
 
 WORKDIR /app
 COPY --chown=openhands:openhands --chmod=770 enterprise .

enterprise/downgrade_migrated_users.py

@@ -0,0 +1,207 @@
+#!/usr/bin/env python
+"""
+This script can be removed once orgs is established - probably after Feb 15 2026
+
+Downgrade script for migrated users.
+
+This script identifies users who have been migrated (already_migrated=True)
+and reverts them back to the pre-migration state.
+
+Usage:
+    # Dry run - just list the users that would be downgraded
+    python downgrade_migrated_users.py --dry-run
+
+    # Downgrade a specific user by their keycloak_user_id
+    python downgrade_migrated_users.py --user-id <user_id>
+
+    # Downgrade all migrated users (with confirmation)
+    python downgrade_migrated_users.py --all
+
+    # Downgrade all migrated users without confirmation (dangerous!)
+    python downgrade_migrated_users.py --all --no-confirm
+"""
+
+import argparse
+import asyncio
+import sys
+
+# Add the enterprise directory to the path
+sys.path.insert(0, '/workspace/project/OpenHands/enterprise')
+
+from server.logger import logger
+from sqlalchemy import select, text
+from storage.database import session_maker
+from storage.user_settings import UserSettings
+from storage.user_store import UserStore
+
+
+def get_migrated_users() -> list[str]:
+    """Get list of keycloak_user_ids for users who have been migrated.
+
+    This includes:
+    1. Users with already_migrated=True in user_settings (migrated users)
+    2. Users in the 'user' table who don't have a user_settings entry (new sign-ups)
+    """
+    with session_maker() as session:
+        # Get users from user_settings with already_migrated=True
+        migrated_result = session.execute(
+            select(UserSettings.keycloak_user_id).where(
+                UserSettings.already_migrated.is_(True)
+            )
+        )
+        migrated_users = {row[0] for row in migrated_result.fetchall() if row[0]}
+
+        # Get users from the 'user' table (new sign-ups won't have user_settings)
+        # These are users who signed up after the migration was deployed
+        new_signup_result = session.execute(
+            text("""
+                SELECT CAST(u.id AS VARCHAR)
+                FROM "user" u
+                WHERE NOT EXISTS (
+                    SELECT 1 FROM user_settings us
+                    WHERE us.keycloak_user_id = CAST(u.id AS VARCHAR)
+                )
+            """)
+        )
+        new_signups = {row[0] for row in new_signup_result.fetchall() if row[0]}
+
+        # Combine both sets
+        all_users = migrated_users | new_signups
+        return list(all_users)
+
+
+async def downgrade_user(user_id: str) -> bool:
+    """Downgrade a single user.
+
+    Args:
+        user_id: The keycloak_user_id to downgrade
+
+    Returns:
+        True if successful, False otherwise
+    """
+    try:
+        result = await UserStore.downgrade_user(user_id)
+        if result:
+            print(f'✓ Successfully downgraded user: {user_id}')
+            return True
+        else:
+            print(f'✗ Failed to downgrade user: {user_id}')
+            return False
+    except Exception as e:
+        print(f'✗ Error downgrading user {user_id}: {e}')
+        logger.exception(
+            'downgrade_script:error',
+            extra={'user_id': user_id, 'error': str(e)},
+        )
+        return False
+
+
+async def main():
+    parser = argparse.ArgumentParser(
+        description='Downgrade migrated users back to pre-migration state'
+    )
+    parser.add_argument(
+        '--dry-run',
+        action='store_true',
+        help='Just list users that would be downgraded, without making changes',
+    )
+    parser.add_argument(
+        '--user-id',
+        type=str,
+        help='Downgrade a specific user by keycloak_user_id',
+    )
+    parser.add_argument(
+        '--all',
+        action='store_true',
+        help='Downgrade all migrated users',
+    )
+    parser.add_argument(
+        '--no-confirm',
+        action='store_true',
+        help='Skip confirmation prompt (use with caution!)',
+    )
+
+    args = parser.parse_args()
+
+    # Get list of migrated users
+    migrated_users = get_migrated_users()
+    print(f'\nFound {len(migrated_users)} migrated user(s).')
+
+    if args.dry_run:
+        print('\n--- DRY RUN MODE ---')
+        print('The following users would be downgraded:')
+        for user_id in migrated_users:
+            print(f'  - {user_id}')
+        print('\nNo changes were made.')
+        return
+
+    if args.user_id:
+        # Downgrade a specific user
+        if args.user_id not in migrated_users:
+            print(f'\nUser {args.user_id} is not in the migrated users list.')
+            print('Either the user was not migrated, or the user_id is incorrect.')
+            return
+
+        print(f'\nDowngrading user: {args.user_id}')
+        if not args.no_confirm:
+            confirm = input('Are you sure? (yes/no): ')
+            if confirm.lower() != 'yes':
+                print('Cancelled.')
+                return
+
+        success = await downgrade_user(args.user_id)
+        if success:
+            print('\nDowngrade completed successfully.')
+        else:
+            print('\nDowngrade failed. Check logs for details.')
+            sys.exit(1)
+
+    elif args.all:
+        # Downgrade all migrated users
+        if not migrated_users:
+            print('\nNo migrated users to downgrade.')
+            return
+
+        print(f'\n⚠️  About to downgrade {len(migrated_users)} user(s).')
+        if not args.no_confirm:
+            print('\nThis will:')
+            print('  - Revert LiteLLM team/user budget settings')
+            print('  - Delete organization entries')
+            print('  - Delete user entries in the new schema')
+            print('  - Reset the already_migrated flag')
+            print('\nUsers to downgrade:')
+            for user_id in migrated_users[:10]:  # Show first 10
+                print(f'  - {user_id}')
+            if len(migrated_users) > 10:
+                print(f'  ... and {len(migrated_users) - 10} more')
+
+            confirm = input('\nType "yes" to proceed: ')
+            if confirm.lower() != 'yes':
+                print('Cancelled.')
+                return
+
+        print('\nStarting downgrade...\n')
+        success_count = 0
+        fail_count = 0
+
+        for user_id in migrated_users:
+            success = await downgrade_user(user_id)
+            if success:
+                success_count += 1
+            else:
+                fail_count += 1
+
+        print('\n--- Summary ---')
+        print(f'Successful: {success_count}')
+        print(f'Failed: {fail_count}')
+
+        if fail_count > 0:
+            sys.exit(1)
+
+    else:
+        parser.print_help()
+        print('\nPlease specify --dry-run, --user-id, or --all')
+
+
+if __name__ == '__main__':
+    asyncio.run(main())

enterprise/integrations/github/github_manager.py

@@ -26,12 +26,14 @@ from integrations.utils import (
 from integrations.v1_utils import get_saas_user_auth
 from jinja2 import Environment, FileSystemLoader
 from pydantic import SecretStr
+from server.auth.auth_error import ExpiredError
 from server.auth.constants import GITHUB_APP_CLIENT_ID, GITHUB_APP_PRIVATE_KEY
 from server.auth.token_manager import TokenManager
 from server.utils.conversation_callback_utils import register_callback_processor
 
 from openhands.core.logger import openhands_logger as logger
 from openhands.integrations.provider import ProviderToken, ProviderType
+from openhands.integrations.service_types import AuthenticationError
 from openhands.server.types import (
     LLMAuthenticationError,
     MissingSettingsError,
@@ -143,11 +145,7 @@ class GithubManager(Manager):
         ).get('body', ''):
             return False
 
-        if GithubFactory.is_eligible_for_conversation_starter(
-            message
-        ) and self._user_has_write_access_to_repo(installation_id, repo_name, username):
-            await GithubFactory.trigger_conversation_starter(message)
-
+        # Check event types before making expensive API calls (e.g., _user_has_write_access_to_repo)
         if not (
             GithubFactory.is_labeled_issue(message)
             or GithubFactory.is_issue_comment(message)
@@ -157,8 +155,17 @@ class GithubManager(Manager):
             return False
 
         logger.info(f'[GitHub] Checking permissions for {username} in {repo_name}')
+        user_has_write_access = self._user_has_write_access_to_repo(
+            installation_id, repo_name, username
+        )
 
-        return self._user_has_write_access_to_repo(installation_id, repo_name, username)
+        if (
+            GithubFactory.is_eligible_for_conversation_starter(message)
+            and user_has_write_access
+        ):
+            await GithubFactory.trigger_conversation_starter(message)
+
+        return user_has_write_access
 
     async def receive_message(self, message: Message):
         self._confirm_incoming_source_type(message)
@@ -347,7 +354,7 @@ class GithubManager(Manager):
 
                 msg_info = f'@{user_info.username} please set a valid LLM API key in [OpenHands Cloud]({HOST_URL}) before starting a job.'
 
-            except SessionExpiredError as e:
+            except (AuthenticationError, ExpiredError, SessionExpiredError) as e:
                 logger.warning(
                     f'[GitHub] Session expired for user {user_info.username}: {str(e)}'
                 )

enterprise/integrations/gitlab/webhook_installation.py

@@ -167,17 +167,15 @@ async def install_webhook_on_resource(
         scopes=SCOPES,
     )
 
-    logger.info(
-        'Creating new webhook',
-        extra={
-            'webhook_id': webhook_id,
-            'status': status,
-            'resource_id': resource_id,
-            'resource_type': resource_type,
-        },
-    )
+    log_extra = {
+        'webhook_id': webhook_id,
+        'status': status,
+        'resource_id': resource_id,
+        'resource_type': resource_type,
+    }
 
     if status == WebhookStatus.RATE_LIMITED:
+        logger.warning('Rate limited while creating webhook', extra=log_extra)
         raise BreakLoopException()
 
     if webhook_id:
@@ -191,9 +189,8 @@ async def install_webhook_on_resource(
                 'webhook_uuid': webhook_uuid,  # required to identify which webhook installation is sending payload
             },
         )
-
-        logger.info(
-            f'Installed webhook for {webhook.user_id} on {resource_type}:{resource_id}'
-        )
+        logger.info('Created new webhook', extra=log_extra)
+    else:
+        logger.error('Failed to create webhook', extra=log_extra)
 
     return webhook_id, status

enterprise/integrations/resolver_context.py

@@ -1,6 +1,6 @@
 from openhands.app_server.user.user_context import UserContext
 from openhands.app_server.user.user_models import UserInfo
-from openhands.integrations.provider import PROVIDER_TOKEN_TYPE
+from openhands.integrations.provider import PROVIDER_TOKEN_TYPE, ProviderHandler
 from openhands.integrations.service_types import ProviderType
 from openhands.sdk.secret import SecretSource, StaticSecret
 from openhands.server.user_auth.user_auth import UserAuth
@@ -14,6 +14,7 @@ class ResolverUserContext(UserContext):
         saas_user_auth: UserAuth,
     ):
         self.saas_user_auth = saas_user_auth
+        self._provider_handler: ProviderHandler | None = None
 
     async def get_user_id(self) -> str | None:
         return await self.saas_user_auth.get_user_id()
@@ -29,12 +30,26 @@ class ResolverUserContext(UserContext):
 
         return UserInfo(id=user_id)
 
+    async def _get_provider_handler(self) -> ProviderHandler:
+        """Get or create a ProviderHandler for git operations."""
+        if self._provider_handler is None:
+            provider_tokens = await self.saas_user_auth.get_provider_tokens()
+            if provider_tokens is None:
+                raise ValueError('No provider tokens available')
+            user_id = await self.saas_user_auth.get_user_id()
+            self._provider_handler = ProviderHandler(
+                provider_tokens=provider_tokens, external_auth_id=user_id
+            )
+        return self._provider_handler
+
     async def get_authenticated_git_url(
         self, repository: str, is_optional: bool = False
     ) -> str:
-        # This would need to be implemented based on the git provider tokens
-        # For now, return a basic HTTPS URL
-        return f'https://github.com/{repository}.git'
+        provider_handler = await self._get_provider_handler()
+        url = await provider_handler.get_authenticated_git_url(
+            repository, is_optional=is_optional
+        )
+        return url
 
     async def get_latest_token(self, provider_type: ProviderType) -> str | None:
         # Return the appropriate token string from git_provider_tokens

enterprise/migrations/env.py

@@ -1,10 +1,15 @@
+import logging
 import os
 from logging.config import fileConfig
 
-from alembic import context
-from google.cloud.sql.connector import Connector
-from sqlalchemy import create_engine
-from storage.base import Base
+# Suppress alembic.runtime.plugins INFO logs during import to prevent non-JSON logs in production
+# These plugin setup messages would otherwise appear before logging is configured
+logging.getLogger('alembic.runtime.plugins').setLevel(logging.WARNING)
+
+from alembic import context  # noqa: E402
+from google.cloud.sql.connector import Connector  # noqa: E402
+from sqlalchemy import create_engine  # noqa: E402
+from storage.base import Base  # noqa: E402
 
 target_metadata = Base.metadata
 

enterprise/migrations/versions/090_add_git_user_fields_to_user.py

@@ -0,0 +1,28 @@
+"""Add git_user_name and git_user_email columns to user table.
+
+Revision ID: 090
+Revises: 089
+Create Date: 2025-01-22
+"""
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = '090'
+down_revision = '089'
+
+
+def upgrade() -> None:
+    op.add_column(
+        'user',
+        sa.Column('git_user_name', sa.String, nullable=True),
+    )
+    op.add_column(
+        'user',
+        sa.Column('git_user_email', sa.String, nullable=True),
+    )
+
+
+def downgrade() -> None:
+    op.drop_column('user', 'git_user_email')
+    op.drop_column('user', 'git_user_name')

enterprise/migrations/versions/091_add_byor_export_enabled_flag.py

@@ -0,0 +1,46 @@
+"""Add byor_export_enabled flag to org table.
+
+Revision ID: 091
+Revises: 090
+Create Date: 2025-01-15 00:00:00.000000
+
+"""
+
+from typing import Sequence, Union
+
+import sqlalchemy as sa
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision: str = '091'
+down_revision: Union[str, None] = '090'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # Add byor_export_enabled column to org table with default false
+    op.add_column(
+        'org',
+        sa.Column(
+            'byor_export_enabled',
+            sa.Boolean,
+            nullable=False,
+            server_default=sa.text('false'),
+        ),
+    )
+
+    # Set byor_export_enabled to true for orgs that have completed billing sessions
+    op.execute(
+        sa.text("""
+            UPDATE org SET byor_export_enabled = TRUE
+            WHERE id IN (
+                SELECT DISTINCT org_id FROM billing_sessions
+                WHERE status = 'completed' AND org_id IS NOT NULL
+            )
+        """)
+    )
+
+
+def downgrade() -> None:
+    op.drop_column('org', 'byor_export_enabled')

enterprise/migrations/versions/092_rename_user_role_to_member.py

@@ -0,0 +1,29 @@
+"""Rename 'user' role to 'member' in role table.
+
+Revision ID: 092
+Revises: 091
+Create Date: 2025-02-12 00:00:00.000000
+
+"""
+
+from typing import Sequence, Union
+
+import sqlalchemy as sa
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision: str = '092'
+down_revision: Union[str, None] = '091'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # Rename 'user' role to 'member' for clarity
+    # This avoids confusion between the 'user' role and the 'user' entity/account
+    op.execute(sa.text("UPDATE role SET name = 'member' WHERE name = 'user'"))
+
+
+def downgrade() -> None:
+    # Revert 'member' role back to 'user'
+    op.execute(sa.text("UPDATE role SET name = 'user' WHERE name = 'member'"))

enterprise/poetry.lock

@@ -6102,14 +6102,14 @@ llama = ["llama-index (>=0.12.29,<0.13.0)", "llama-index-core (>=0.12.29,<0.13.0
 
 [[package]]
 name = "openhands-agent-server"
-version = "1.9.0"
+version = "1.11.4"
 description = "OpenHands Agent Server - REST/WebSocket interface for OpenHands AI Agent"
 optional = false
 python-versions = ">=3.12"
 groups = ["main"]
 files = [
-    {file = "openhands_agent_server-1.9.0-py3-none-any.whl", hash = "sha256:44b65fac5bb831541eb2e8726afb2682bde4816b4c6c90be9ad3cafd3dbcf971"},
-    {file = "openhands_agent_server-1.9.0.tar.gz", hash = "sha256:ac41a948acf64ed661a9f383c293c305176f92bd12e6fc6362f5414cb7874ee1"},
+    {file = "openhands_agent_server-1.11.4-py3-none-any.whl", hash = "sha256:739bdb774dbfcd23d6e87ee6ee32bc0999f22300037506b6dd33e9ea67fa5c2a"},
+    {file = "openhands_agent_server-1.11.4.tar.gz", hash = "sha256:41247f7022a046eb50ca3b552bc6d12bfa9776e1bd27d0989da91b9f7ac77ca2"},
 ]
 
 [package.dependencies]
@@ -6126,7 +6126,7 @@ wsproto = ">=1.2.0"
 
 [[package]]
 name = "openhands-ai"
-version = "1.2.1"
+version = "1.3.0"
 description = "OpenHands: Code Less, Make More"
 optional = false
 python-versions = "^3.12,<3.14"
@@ -6168,9 +6168,9 @@ memory-profiler = ">=0.61"
 numpy = "*"
 openai = "2.8"
 openhands-aci = "0.3.2"
-openhands-agent-server = "1.9"
-openhands-sdk = "1.9"
-openhands-tools = "1.9"
+openhands-agent-server = "1.11.4"
+openhands-sdk = "1.11.4"
+openhands-tools = "1.11.4"
 opentelemetry-api = ">=1.33.1"
 opentelemetry-exporter-otlp-proto-grpc = ">=1.33.1"
 pathspec = ">=0.12.1"
@@ -6225,14 +6225,14 @@ url = ".."
 
 [[package]]
 name = "openhands-sdk"
-version = "1.9.0"
+version = "1.11.4"
 description = "OpenHands SDK - Core functionality for building AI agents"
 optional = false
 python-versions = ">=3.12"
 groups = ["main"]
 files = [
-    {file = "openhands_sdk-1.9.0-py3-none-any.whl", hash = "sha256:b427d8b9e587a5360c7d61742c290601998557e9b38b1c9e11a297659812c00d"},
-    {file = "openhands_sdk-1.9.0.tar.gz", hash = "sha256:70048888fd4fbe44a86c35c402bbb99d30cf0cba50579ee1a8e3f43e05154150"},
+    {file = "openhands_sdk-1.11.4-py3-none-any.whl", hash = "sha256:9f4607c5d94b56fbcd533207026ee892779dd50e29bce79277ff82454a4f76d5"},
+    {file = "openhands_sdk-1.11.4.tar.gz", hash = "sha256:4088744f6b8856eeab22d3bc17e47d1736ea7ced945c2fa126bd7d48c14bb313"},
 ]
 
 [package.dependencies]
@@ -6253,14 +6253,14 @@ boto3 = ["boto3 (>=1.35.0)"]
 
 [[package]]
 name = "openhands-tools"
-version = "1.9.0"
+version = "1.11.4"
 description = "OpenHands Tools - Runtime tools for AI agents"
 optional = false
 python-versions = ">=3.12"
 groups = ["main"]
 files = [
-    {file = "openhands_tools-1.9.0-py3-none-any.whl", hash = "sha256:8becde0e913a31babb41eb93a8c10bf41d87ca1febd07bc958839c3583655305"},
-    {file = "openhands_tools-1.9.0.tar.gz", hash = "sha256:d45f5f5210cb2bbcd8ab5f3a32051db1a532d0ec07cd306105f95cde42cf67f2"},
+    {file = "openhands_tools-1.11.4-py3-none-any.whl", hash = "sha256:efd721b73e87a0dac69171a76931363fa59fcde98107ca86081ee7bf0253673a"},
+    {file = "openhands_tools-1.11.4.tar.gz", hash = "sha256:80671b1ea8c85a5247a75ea2340ae31d76363e9c723b104699a9a77e66d2043c"},
 ]
 
 [package.dependencies]
@@ -6851,103 +6851,103 @@ scramp = ">=1.4.5"
 
 [[package]]
 name = "pillow"
-version = "12.1.0"
+version = "12.1.1"
 description = "Python Imaging Library (fork)"
 optional = false
 python-versions = ">=3.10"
 groups = ["main", "test"]
 files = [
-    {file = "pillow-12.1.0-cp310-cp310-macosx_10_10_x86_64.whl", hash = "sha256:fb125d860738a09d363a88daa0f59c4533529a90e564785e20fe875b200b6dbd"},
-    {file = "pillow-12.1.0-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:cad302dc10fac357d3467a74a9561c90609768a6f73a1923b0fd851b6486f8b0"},
-    {file = "pillow-12.1.0-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:a40905599d8079e09f25027423aed94f2823adaf2868940de991e53a449e14a8"},
-    {file = "pillow-12.1.0-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:92a7fe4225365c5e3a8e598982269c6d6698d3e783b3b1ae979e7819f9cd55c1"},
-    {file = "pillow-12.1.0-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f10c98f49227ed8383d28174ee95155a675c4ed7f85e2e573b04414f7e371bda"},
-    {file = "pillow-12.1.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:8637e29d13f478bc4f153d8daa9ffb16455f0a6cb287da1b432fdad2bfbd66c7"},
-    {file = "pillow-12.1.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:21e686a21078b0f9cb8c8a961d99e6a4ddb88e0fc5ea6e130172ddddc2e5221a"},
-    {file = "pillow-12.1.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:2415373395a831f53933c23ce051021e79c8cd7979822d8cc478547a3f4da8ef"},
-    {file = "pillow-12.1.0-cp310-cp310-win32.whl", hash = "sha256:e75d3dba8fc1ddfec0cd752108f93b83b4f8d6ab40e524a95d35f016b9683b09"},
-    {file = "pillow-12.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:64efdf00c09e31efd754448a383ea241f55a994fd079866b92d2bbff598aad91"},
-    {file = "pillow-12.1.0-cp310-cp310-win_arm64.whl", hash = "sha256:f188028b5af6b8fb2e9a76ac0f841a575bd1bd396e46ef0840d9b88a48fdbcea"},
-    {file = "pillow-12.1.0-cp311-cp311-macosx_10_10_x86_64.whl", hash = "sha256:a83e0850cb8f5ac975291ebfc4170ba481f41a28065277f7f735c202cd8e0af3"},
-    {file = "pillow-12.1.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:b6e53e82ec2db0717eabb276aa56cf4e500c9a7cec2c2e189b55c24f65a3e8c0"},
-    {file = "pillow-12.1.0-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:40a8e3b9e8773876d6e30daed22f016509e3987bab61b3b7fe309d7019a87451"},
-    {file = "pillow-12.1.0-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:800429ac32c9b72909c671aaf17ecd13110f823ddb7db4dfef412a5587c2c24e"},
-    {file = "pillow-12.1.0-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:0b022eaaf709541b391ee069f0022ee5b36c709df71986e3f7be312e46f42c84"},
-    {file = "pillow-12.1.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:1f345e7bc9d7f368887c712aa5054558bad44d2a301ddf9248599f4161abc7c0"},
-    {file = "pillow-12.1.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:d70347c8a5b7ccd803ec0c85c8709f036e6348f1e6a5bf048ecd9c64d3550b8b"},
-    {file = "pillow-12.1.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:1fcc52d86ce7a34fd17cb04e87cfdb164648a3662a6f20565910a99653d66c18"},
-    {file = "pillow-12.1.0-cp311-cp311-win32.whl", hash = "sha256:3ffaa2f0659e2f740473bcf03c702c39a8d4b2b7ffc629052028764324842c64"},
-    {file = "pillow-12.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:806f3987ffe10e867bab0ddad45df1148a2b98221798457fa097ad85d6e8bc75"},
-    {file = "pillow-12.1.0-cp311-cp311-win_arm64.whl", hash = "sha256:9f5fefaca968e700ad1a4a9de98bf0869a94e397fe3524c4c9450c1445252304"},
-    {file = "pillow-12.1.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:a332ac4ccb84b6dde65dbace8431f3af08874bf9770719d32a635c4ef411b18b"},
-    {file = "pillow-12.1.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:907bfa8a9cb790748a9aa4513e37c88c59660da3bcfffbd24a7d9e6abf224551"},
-    {file = "pillow-12.1.0-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:efdc140e7b63b8f739d09a99033aa430accce485ff78e6d311973a67b6bf3208"},
-    {file = "pillow-12.1.0-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:bef9768cab184e7ae6e559c032e95ba8d07b3023c289f79a2bd36e8bf85605a5"},
-    {file = "pillow-12.1.0-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:742aea052cf5ab5034a53c3846165bc3ce88d7c38e954120db0ab867ca242661"},
-    {file = "pillow-12.1.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a6dfc2af5b082b635af6e08e0d1f9f1c4e04d17d4e2ca0ef96131e85eda6eb17"},
-    {file = "pillow-12.1.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:609e89d9f90b581c8d16358c9087df76024cf058fa693dd3e1e1620823f39670"},
-    {file = "pillow-12.1.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:43b4899cfd091a9693a1278c4982f3e50f7fb7cff5153b05174b4afc9593b616"},
-    {file = "pillow-12.1.0-cp312-cp312-win32.whl", hash = "sha256:aa0c9cc0b82b14766a99fbe6084409972266e82f459821cd26997a488a7261a7"},
-    {file = "pillow-12.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:d70534cea9e7966169ad29a903b99fc507e932069a881d0965a1a84bb57f6c6d"},
-    {file = "pillow-12.1.0-cp312-cp312-win_arm64.whl", hash = "sha256:65b80c1ee7e14a87d6a068dd3b0aea268ffcabfe0498d38661b00c5b4b22e74c"},
-    {file = "pillow-12.1.0-cp313-cp313-ios_13_0_arm64_iphoneos.whl", hash = "sha256:7b5dd7cbae20285cdb597b10eb5a2c13aa9de6cde9bb64a3c1317427b1db1ae1"},
-    {file = "pillow-12.1.0-cp313-cp313-ios_13_0_arm64_iphonesimulator.whl", hash = "sha256:29a4cef9cb672363926f0470afc516dbf7305a14d8c54f7abbb5c199cd8f8179"},
-    {file = "pillow-12.1.0-cp313-cp313-ios_13_0_x86_64_iphonesimulator.whl", hash = "sha256:681088909d7e8fa9e31b9799aaa59ba5234c58e5e4f1951b4c4d1082a2e980e0"},
-    {file = "pillow-12.1.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:983976c2ab753166dc66d36af6e8ec15bb511e4a25856e2227e5f7e00a160587"},
-    {file = "pillow-12.1.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:db44d5c160a90df2d24a24760bbd37607d53da0b34fb546c4c232af7192298ac"},
-    {file = "pillow-12.1.0-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:6b7a9d1db5dad90e2991645874f708e87d9a3c370c243c2d7684d28f7e133e6b"},
-    {file = "pillow-12.1.0-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:6258f3260986990ba2fa8a874f8b6e808cf5abb51a94015ca3dc3c68aa4f30ea"},
-    {file = "pillow-12.1.0-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:e115c15e3bc727b1ca3e641a909f77f8ca72a64fff150f666fcc85e57701c26c"},
-    {file = "pillow-12.1.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:6741e6f3074a35e47c77b23a4e4f2d90db3ed905cb1c5e6e0d49bff2045632bc"},
-    {file = "pillow-12.1.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:935b9d1aed48fcfb3f838caac506f38e29621b44ccc4f8a64d575cb1b2a88644"},
-    {file = "pillow-12.1.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:5fee4c04aad8932da9f8f710af2c1a15a83582cfb884152a9caa79d4efcdbf9c"},
-    {file = "pillow-12.1.0-cp313-cp313-win32.whl", hash = "sha256:a786bf667724d84aa29b5db1c61b7bfdde380202aaca12c3461afd6b71743171"},
-    {file = "pillow-12.1.0-cp313-cp313-win_amd64.whl", hash = "sha256:461f9dfdafa394c59cd6d818bdfdbab4028b83b02caadaff0ffd433faf4c9a7a"},
-    {file = "pillow-12.1.0-cp313-cp313-win_arm64.whl", hash = "sha256:9212d6b86917a2300669511ed094a9406888362e085f2431a7da985a6b124f45"},
-    {file = "pillow-12.1.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:00162e9ca6d22b7c3ee8e61faa3c3253cd19b6a37f126cad04f2f88b306f557d"},
-    {file = "pillow-12.1.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:7d6daa89a00b58c37cb1747ec9fb7ac3bc5ffd5949f5888657dfddde6d1312e0"},
-    {file = "pillow-12.1.0-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:e2479c7f02f9d505682dc47df8c0ea1fc5e264c4d1629a5d63fe3e2334b89554"},
-    {file = "pillow-12.1.0-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:f188d580bd870cda1e15183790d1cc2fa78f666e76077d103edf048eed9c356e"},
-    {file = "pillow-12.1.0-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:0fde7ec5538ab5095cc02df38ee99b0443ff0e1c847a045554cf5f9af1f4aa82"},
-    {file = "pillow-12.1.0-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0ed07dca4a8464bada6139ab38f5382f83e5f111698caf3191cb8dbf27d908b4"},
-    {file = "pillow-12.1.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:f45bd71d1fa5e5749587613037b172e0b3b23159d1c00ef2fc920da6f470e6f0"},
-    {file = "pillow-12.1.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:277518bf4fe74aa91489e1b20577473b19ee70fb97c374aa50830b279f25841b"},
-    {file = "pillow-12.1.0-cp313-cp313t-win32.whl", hash = "sha256:7315f9137087c4e0ee73a761b163fc9aa3b19f5f606a7fc08d83fd3e4379af65"},
-    {file = "pillow-12.1.0-cp313-cp313t-win_amd64.whl", hash = "sha256:0ddedfaa8b5f0b4ffbc2fa87b556dc59f6bb4ecb14a53b33f9189713ae8053c0"},
-    {file = "pillow-12.1.0-cp313-cp313t-win_arm64.whl", hash = "sha256:80941e6d573197a0c28f394753de529bb436b1ca990ed6e765cf42426abc39f8"},
-    {file = "pillow-12.1.0-cp314-cp314-ios_13_0_arm64_iphoneos.whl", hash = "sha256:5cb7bc1966d031aec37ddb9dcf15c2da5b2e9f7cc3ca7c54473a20a927e1eb91"},
-    {file = "pillow-12.1.0-cp314-cp314-ios_13_0_arm64_iphonesimulator.whl", hash = "sha256:97e9993d5ed946aba26baf9c1e8cf18adbab584b99f452ee72f7ee8acb882796"},
-    {file = "pillow-12.1.0-cp314-cp314-ios_13_0_x86_64_iphonesimulator.whl", hash = "sha256:414b9a78e14ffeb98128863314e62c3f24b8a86081066625700b7985b3f529bd"},
-    {file = "pillow-12.1.0-cp314-cp314-macosx_10_15_x86_64.whl", hash = "sha256:e6bdb408f7c9dd2a5ff2b14a3b0bb6d4deb29fb9961e6eb3ae2031ae9a5cec13"},
-    {file = "pillow-12.1.0-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:3413c2ae377550f5487991d444428f1a8ae92784aac79caa8b1e3b89b175f77e"},
-    {file = "pillow-12.1.0-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:e5dcbe95016e88437ecf33544ba5db21ef1b8dd6e1b434a2cb2a3d605299e643"},
-    {file = "pillow-12.1.0-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:d0a7735df32ccbcc98b98a1ac785cc4b19b580be1bdf0aeb5c03223220ea09d5"},
-    {file = "pillow-12.1.0-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:0c27407a2d1b96774cbc4a7594129cc027339fd800cd081e44497722ea1179de"},
-    {file = "pillow-12.1.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:15c794d74303828eaa957ff8070846d0efe8c630901a1c753fdc63850e19ecd9"},
-    {file = "pillow-12.1.0-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:c990547452ee2800d8506c4150280757f88532f3de2a58e3022e9b179107862a"},
-    {file = "pillow-12.1.0-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:b63e13dd27da389ed9475b3d28510f0f954bca0041e8e551b2a4eb1eab56a39a"},
-    {file = "pillow-12.1.0-cp314-cp314-win32.whl", hash = "sha256:1a949604f73eb07a8adab38c4fe50791f9919344398bdc8ac6b307f755fc7030"},
-    {file = "pillow-12.1.0-cp314-cp314-win_amd64.whl", hash = "sha256:4f9f6a650743f0ddee5593ac9e954ba1bdbc5e150bc066586d4f26127853ab94"},
-    {file = "pillow-12.1.0-cp314-cp314-win_arm64.whl", hash = "sha256:808b99604f7873c800c4840f55ff389936ef1948e4e87645eaf3fccbc8477ac4"},
-    {file = "pillow-12.1.0-cp314-cp314t-macosx_10_15_x86_64.whl", hash = "sha256:bc11908616c8a283cf7d664f77411a5ed2a02009b0097ff8abbba5e79128ccf2"},
-    {file = "pillow-12.1.0-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:896866d2d436563fa2a43a9d72f417874f16b5545955c54a64941e87c1376c61"},
-    {file = "pillow-12.1.0-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:8e178e3e99d3c0ea8fc64b88447f7cac8ccf058af422a6cedc690d0eadd98c51"},
-    {file = "pillow-12.1.0-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:079af2fb0c599c2ec144ba2c02766d1b55498e373b3ac64687e43849fbbef5bc"},
-    {file = "pillow-12.1.0-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:bdec5e43377761c5dbca620efb69a77f6855c5a379e32ac5b158f54c84212b14"},
-    {file = "pillow-12.1.0-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:565c986f4b45c020f5421a4cea13ef294dde9509a8577f29b2fc5edc7587fff8"},
-    {file = "pillow-12.1.0-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:43aca0a55ce1eefc0aefa6253661cb54571857b1a7b2964bd8a1e3ef4b729924"},
-    {file = "pillow-12.1.0-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:0deedf2ea233722476b3a81e8cdfbad786f7adbed5d848469fa59fe52396e4ef"},
-    {file = "pillow-12.1.0-cp314-cp314t-win32.whl", hash = "sha256:b17fbdbe01c196e7e159aacb889e091f28e61020a8abeac07b68079b6e626988"},
-    {file = "pillow-12.1.0-cp314-cp314t-win_amd64.whl", hash = "sha256:27b9baecb428899db6c0de572d6d305cfaf38ca1596b5c0542a5182e3e74e8c6"},
-    {file = "pillow-12.1.0-cp314-cp314t-win_arm64.whl", hash = "sha256:f61333d817698bdcdd0f9d7793e365ac3d2a21c1f1eb02b32ad6aefb8d8ea831"},
-    {file = "pillow-12.1.0-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:ca94b6aac0d7af2a10ba08c0f888b3d5114439b6b3ef39968378723622fed377"},
-    {file = "pillow-12.1.0-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:351889afef0f485b84078ea40fe33727a0492b9af3904661b0abbafee0355b72"},
-    {file = "pillow-12.1.0-pp311-pypy311_pp73-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:bb0984b30e973f7e2884362b7d23d0a348c7143ee559f38ef3eaab640144204c"},
-    {file = "pillow-12.1.0-pp311-pypy311_pp73-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:84cabc7095dd535ca934d57e9ce2a72ffd216e435a84acb06b2277b1de2689bd"},
-    {file = "pillow-12.1.0-pp311-pypy311_pp73-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:53d8b764726d3af1a138dd353116f774e3862ec7e3794e0c8781e30db0f35dfc"},
-    {file = "pillow-12.1.0-pp311-pypy311_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5da841d81b1a05ef940a8567da92decaa15bc4d7dedb540a8c219ad83d91808a"},
-    {file = "pillow-12.1.0-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:75af0b4c229ac519b155028fa1be632d812a519abba9b46b20e50c6caa184f19"},
-    {file = "pillow-12.1.0.tar.gz", hash = "sha256:5c5ae0a06e9ea030ab786b0251b32c7e4ce10e58d983c0d5c56029455180b5b9"},
+    {file = "pillow-12.1.1-cp310-cp310-macosx_10_10_x86_64.whl", hash = "sha256:1f1625b72740fdda5d77b4def688eb8fd6490975d06b909fd19f13f391e077e0"},
+    {file = "pillow-12.1.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:178aa072084bd88ec759052feca8e56cbb14a60b39322b99a049e58090479713"},
+    {file = "pillow-12.1.1-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:b66e95d05ba806247aaa1561f080abc7975daf715c30780ff92a20e4ec546e1b"},
+    {file = "pillow-12.1.1-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:89c7e895002bbe49cdc5426150377cbbc04767d7547ed145473f496dfa40408b"},
+    {file = "pillow-12.1.1-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:3a5cbdcddad0af3da87cb16b60d23648bc3b51967eb07223e9fed77a82b457c4"},
+    {file = "pillow-12.1.1-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:9f51079765661884a486727f0729d29054242f74b46186026582b4e4769918e4"},
+    {file = "pillow-12.1.1-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:99c1506ea77c11531d75e3a412832a13a71c7ebc8192ab9e4b2e355555920e3e"},
+    {file = "pillow-12.1.1-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:36341d06738a9f66c8287cf8b876d24b18db9bd8740fa0672c74e259ad408cff"},
+    {file = "pillow-12.1.1-cp310-cp310-win32.whl", hash = "sha256:6c52f062424c523d6c4db85518774cc3d50f5539dd6eed32b8f6229b26f24d40"},
+    {file = "pillow-12.1.1-cp310-cp310-win_amd64.whl", hash = "sha256:c6008de247150668a705a6338156efb92334113421ceecf7438a12c9a12dab23"},
+    {file = "pillow-12.1.1-cp310-cp310-win_arm64.whl", hash = "sha256:1a9b0ee305220b392e1124a764ee4265bd063e54a751a6b62eff69992f457fa9"},
+    {file = "pillow-12.1.1-cp311-cp311-macosx_10_10_x86_64.whl", hash = "sha256:e879bb6cd5c73848ef3b2b48b8af9ff08c5b71ecda8048b7dd22d8a33f60be32"},
+    {file = "pillow-12.1.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:365b10bb9417dd4498c0e3b128018c4a624dc11c7b97d8cc54effe3b096f4c38"},
+    {file = "pillow-12.1.1-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:d4ce8e329c93845720cd2014659ca67eac35f6433fd3050393d85f3ecef0dad5"},
+    {file = "pillow-12.1.1-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:fc354a04072b765eccf2204f588a7a532c9511e8b9c7f900e1b64e3e33487090"},
+    {file = "pillow-12.1.1-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:7e7976bf1910a8116b523b9f9f58bf410f3e8aa330cd9a2bb2953f9266ab49af"},
+    {file = "pillow-12.1.1-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:597bd9c8419bc7c6af5604e55847789b69123bbe25d65cc6ad3012b4f3c98d8b"},
+    {file = "pillow-12.1.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:2c1fc0f2ca5f96a3c8407e41cca26a16e46b21060fe6d5b099d2cb01412222f5"},
+    {file = "pillow-12.1.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:578510d88c6229d735855e1f278aa305270438d36a05031dfaae5067cc8eb04d"},
+    {file = "pillow-12.1.1-cp311-cp311-win32.whl", hash = "sha256:7311c0a0dcadb89b36b7025dfd8326ecfa36964e29913074d47382706e516a7c"},
+    {file = "pillow-12.1.1-cp311-cp311-win_amd64.whl", hash = "sha256:fbfa2a7c10cc2623f412753cddf391c7f971c52ca40a3f65dc5039b2939e8563"},
+    {file = "pillow-12.1.1-cp311-cp311-win_arm64.whl", hash = "sha256:b81b5e3511211631b3f672a595e3221252c90af017e399056d0faabb9538aa80"},
+    {file = "pillow-12.1.1-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:ab323b787d6e18b3d91a72fc99b1a2c28651e4358749842b8f8dfacd28ef2052"},
+    {file = "pillow-12.1.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:adebb5bee0f0af4909c30db0d890c773d1a92ffe83da908e2e9e720f8edf3984"},
+    {file = "pillow-12.1.1-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:bb66b7cc26f50977108790e2456b7921e773f23db5630261102233eb355a3b79"},
+    {file = "pillow-12.1.1-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:aee2810642b2898bb187ced9b349e95d2a7272930796e022efaf12e99dccd293"},
+    {file = "pillow-12.1.1-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:a0b1cd6232e2b618adcc54d9882e4e662a089d5768cd188f7c245b4c8c44a397"},
+    {file = "pillow-12.1.1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7aac39bcf8d4770d089588a2e1dd111cbaa42df5a94be3114222057d68336bd0"},
+    {file = "pillow-12.1.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:ab174cd7d29a62dd139c44bf74b698039328f45cb03b4596c43473a46656b2f3"},
+    {file = "pillow-12.1.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:339ffdcb7cbeaa08221cd401d517d4b1fe7a9ed5d400e4a8039719238620ca35"},
+    {file = "pillow-12.1.1-cp312-cp312-win32.whl", hash = "sha256:5d1f9575a12bed9e9eedd9a4972834b08c97a352bd17955ccdebfeca5913fa0a"},
+    {file = "pillow-12.1.1-cp312-cp312-win_amd64.whl", hash = "sha256:21329ec8c96c6e979cd0dfd29406c40c1d52521a90544463057d2aaa937d66a6"},
+    {file = "pillow-12.1.1-cp312-cp312-win_arm64.whl", hash = "sha256:af9a332e572978f0218686636610555ae3defd1633597be015ed50289a03c523"},
+    {file = "pillow-12.1.1-cp313-cp313-ios_13_0_arm64_iphoneos.whl", hash = "sha256:d242e8ac078781f1de88bf823d70c1a9b3c7950a44cdf4b7c012e22ccbcd8e4e"},
+    {file = "pillow-12.1.1-cp313-cp313-ios_13_0_arm64_iphonesimulator.whl", hash = "sha256:02f84dfad02693676692746df05b89cf25597560db2857363a208e393429f5e9"},
+    {file = "pillow-12.1.1-cp313-cp313-ios_13_0_x86_64_iphonesimulator.whl", hash = "sha256:e65498daf4b583091ccbb2556c7000abf0f3349fcd57ef7adc9a84a394ed29f6"},
+    {file = "pillow-12.1.1-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:6c6db3b84c87d48d0088943bf33440e0c42370b99b1c2a7989216f7b42eede60"},
+    {file = "pillow-12.1.1-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:8b7e5304e34942bf62e15184219a7b5ad4ff7f3bb5cca4d984f37df1a0e1aee2"},
+    {file = "pillow-12.1.1-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:18e5bddd742a44b7e6b1e773ab5db102bd7a94c32555ba656e76d319d19c3850"},
+    {file = "pillow-12.1.1-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:fc44ef1f3de4f45b50ccf9136999d71abb99dca7706bc75d222ed350b9fd2289"},
+    {file = "pillow-12.1.1-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5a8eb7ed8d4198bccbd07058416eeec51686b498e784eda166395a23eb99138e"},
+    {file = "pillow-12.1.1-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:47b94983da0c642de92ced1702c5b6c292a84bd3a8e1d1702ff923f183594717"},
+    {file = "pillow-12.1.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:518a48c2aab7ce596d3bf79d0e275661b846e86e4d0e7dec34712c30fe07f02a"},
+    {file = "pillow-12.1.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:a550ae29b95c6dc13cf69e2c9dc5747f814c54eeb2e32d683e5e93af56caa029"},
+    {file = "pillow-12.1.1-cp313-cp313-win32.whl", hash = "sha256:a003d7422449f6d1e3a34e3dd4110c22148336918ddbfc6a32581cd54b2e0b2b"},
+    {file = "pillow-12.1.1-cp313-cp313-win_amd64.whl", hash = "sha256:344cf1e3dab3be4b1fa08e449323d98a2a3f819ad20f4b22e77a0ede31f0faa1"},
+    {file = "pillow-12.1.1-cp313-cp313-win_arm64.whl", hash = "sha256:5c0dd1636633e7e6a0afe7bf6a51a14992b7f8e60de5789018ebbdfae55b040a"},
+    {file = "pillow-12.1.1-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:0330d233c1a0ead844fc097a7d16c0abff4c12e856c0b325f231820fee1f39da"},
+    {file = "pillow-12.1.1-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:5dae5f21afb91322f2ff791895ddd8889e5e947ff59f71b46041c8ce6db790bc"},
+    {file = "pillow-12.1.1-cp313-cp313t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:2e0c664be47252947d870ac0d327fea7e63985a08794758aa8af5b6cb6ec0c9c"},
+    {file = "pillow-12.1.1-cp313-cp313t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:691ab2ac363b8217f7d31b3497108fb1f50faab2f75dfb03284ec2f217e87bf8"},
+    {file = "pillow-12.1.1-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:e9e8064fb1cc019296958595f6db671fba95209e3ceb0c4734c9baf97de04b20"},
+    {file = "pillow-12.1.1-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:472a8d7ded663e6162dafdf20015c486a7009483ca671cece7a9279b512fcb13"},
+    {file = "pillow-12.1.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:89b54027a766529136a06cfebeecb3a04900397a3590fd252160b888479517bf"},
+    {file = "pillow-12.1.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:86172b0831b82ce4f7877f280055892b31179e1576aa00d0df3bb1bbf8c3e524"},
+    {file = "pillow-12.1.1-cp313-cp313t-win32.whl", hash = "sha256:44ce27545b6efcf0fdbdceb31c9a5bdea9333e664cda58a7e674bb74608b3986"},
+    {file = "pillow-12.1.1-cp313-cp313t-win_amd64.whl", hash = "sha256:a285e3eb7a5a45a2ff504e31f4a8d1b12ef62e84e5411c6804a42197c1cf586c"},
+    {file = "pillow-12.1.1-cp313-cp313t-win_arm64.whl", hash = "sha256:cc7d296b5ea4d29e6570dabeaed58d31c3fea35a633a69679fb03d7664f43fb3"},
+    {file = "pillow-12.1.1-cp314-cp314-ios_13_0_arm64_iphoneos.whl", hash = "sha256:417423db963cb4be8bac3fc1204fe61610f6abeed1580a7a2cbb2fbda20f12af"},
+    {file = "pillow-12.1.1-cp314-cp314-ios_13_0_arm64_iphonesimulator.whl", hash = "sha256:b957b71c6b2387610f556a7eb0828afbe40b4a98036fc0d2acfa5a44a0c2036f"},
+    {file = "pillow-12.1.1-cp314-cp314-ios_13_0_x86_64_iphonesimulator.whl", hash = "sha256:097690ba1f2efdeb165a20469d59d8bb03c55fb6621eb2041a060ae8ea3e9642"},
+    {file = "pillow-12.1.1-cp314-cp314-macosx_10_15_x86_64.whl", hash = "sha256:2815a87ab27848db0321fb78c7f0b2c8649dee134b7f2b80c6a45c6831d75ccd"},
+    {file = "pillow-12.1.1-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:f7ed2c6543bad5a7d5530eb9e78c53132f93dfa44a28492db88b41cdab885202"},
+    {file = "pillow-12.1.1-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:652a2c9ccfb556235b2b501a3a7cf3742148cd22e04b5625c5fe057ea3e3191f"},
+    {file = "pillow-12.1.1-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:d6e4571eedf43af33d0fc233a382a76e849badbccdf1ac438841308652a08e1f"},
+    {file = "pillow-12.1.1-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b574c51cf7d5d62e9be37ba446224b59a2da26dc4c1bb2ecbe936a4fb1a7cb7f"},
+    {file = "pillow-12.1.1-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a37691702ed687799de29a518d63d4682d9016932db66d4e90c345831b02fb4e"},
+    {file = "pillow-12.1.1-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:f95c00d5d6700b2b890479664a06e754974848afaae5e21beb4d83c106923fd0"},
+    {file = "pillow-12.1.1-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:559b38da23606e68681337ad74622c4dbba02254fc9cb4488a305dd5975c7eeb"},
+    {file = "pillow-12.1.1-cp314-cp314-win32.whl", hash = "sha256:03edcc34d688572014ff223c125a3f77fb08091e4607e7745002fc214070b35f"},
+    {file = "pillow-12.1.1-cp314-cp314-win_amd64.whl", hash = "sha256:50480dcd74fa63b8e78235957d302d98d98d82ccbfac4c7e12108ba9ecbdba15"},
+    {file = "pillow-12.1.1-cp314-cp314-win_arm64.whl", hash = "sha256:5cb1785d97b0c3d1d1a16bc1d710c4a0049daefc4935f3a8f31f827f4d3d2e7f"},
+    {file = "pillow-12.1.1-cp314-cp314t-macosx_10_15_x86_64.whl", hash = "sha256:1f90cff8aa76835cba5769f0b3121a22bd4eb9e6884cfe338216e557a9a548b8"},
+    {file = "pillow-12.1.1-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:1f1be78ce9466a7ee64bfda57bdba0f7cc499d9794d518b854816c41bf0aa4e9"},
+    {file = "pillow-12.1.1-cp314-cp314t-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:42fc1f4677106188ad9a55562bbade416f8b55456f522430fadab3cef7cd4e60"},
+    {file = "pillow-12.1.1-cp314-cp314t-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:98edb152429ab62a1818039744d8fbb3ccab98a7c29fc3d5fcef158f3f1f68b7"},
+    {file = "pillow-12.1.1-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d470ab1178551dd17fdba0fef463359c41aaa613cdcd7ff8373f54be629f9f8f"},
+    {file = "pillow-12.1.1-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:6408a7b064595afcab0a49393a413732a35788f2a5092fdc6266952ed67de586"},
+    {file = "pillow-12.1.1-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:5d8c41325b382c07799a3682c1c258469ea2ff97103c53717b7893862d0c98ce"},
+    {file = "pillow-12.1.1-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:c7697918b5be27424e9ce568193efd13d925c4481dd364e43f5dff72d33e10f8"},
+    {file = "pillow-12.1.1-cp314-cp314t-win32.whl", hash = "sha256:d2912fd8114fc5545aa3a4b5576512f64c55a03f3ebcca4c10194d593d43ea36"},
+    {file = "pillow-12.1.1-cp314-cp314t-win_amd64.whl", hash = "sha256:4ceb838d4bd9dab43e06c363cab2eebf63846d6a4aeaea283bbdfd8f1a8ed58b"},
+    {file = "pillow-12.1.1-cp314-cp314t-win_arm64.whl", hash = "sha256:7b03048319bfc6170e93bd60728a1af51d3dd7704935feb228c4d4faab35d334"},
+    {file = "pillow-12.1.1-pp311-pypy311_pp73-macosx_10_15_x86_64.whl", hash = "sha256:600fd103672b925fe62ed08e0d874ea34d692474df6f4bf7ebe148b30f89f39f"},
+    {file = "pillow-12.1.1-pp311-pypy311_pp73-macosx_11_0_arm64.whl", hash = "sha256:665e1b916b043cef294bc54d47bf02d87e13f769bc4bc5fa225a24b3a6c5aca9"},
+    {file = "pillow-12.1.1-pp311-pypy311_pp73-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:495c302af3aad1ca67420ddd5c7bd480c8867ad173528767d906428057a11f0e"},
+    {file = "pillow-12.1.1-pp311-pypy311_pp73-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:8fd420ef0c52c88b5a035a0886f367748c72147b2b8f384c9d12656678dfdfa9"},
+    {file = "pillow-12.1.1-pp311-pypy311_pp73-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f975aa7ef9684ce7e2c18a3aa8f8e2106ce1e46b94ab713d156b2898811651d3"},
+    {file = "pillow-12.1.1-pp311-pypy311_pp73-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:8089c852a56c2966cf18835db62d9b34fef7ba74c726ad943928d494fa7f4735"},
+    {file = "pillow-12.1.1-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:cb9bb857b2d057c6dfc72ac5f3b44836924ba15721882ef103cecb40d002d80e"},
+    {file = "pillow-12.1.1.tar.gz", hash = "sha256:9ad8fa5937ab05218e2b6a4cff30295ad35afd2f83ac592e68c0d871bb0fdbc4"},
 ]
 
 [package.extras]
@@ -14917,4 +14917,4 @@ cffi = ["cffi (>=1.17,<2.0) ; platform_python_implementation != \"PyPy\" and pyt
 [metadata]
 lock-version = "2.1"
 python-versions = "^3.12,<3.14"
-content-hash = "b5cbb1e25176845ac9f95650a802667e2f8be1a536e3e55a9269b5af5a42e3fc"
+content-hash = "1cad6029269393af67155e930c72eae2c03da02e4b3a3699823f6168c14a4218"

enterprise/pyproject.toml

@@ -44,6 +44,12 @@ httpx = "*"
 scikit-learn = "^1.7.0"
 shap = "^0.48.0"
 google-cloud-recaptcha-enterprise = "^1.24.0"
+# Dependencies previously only in Dockerfile, now managed via poetry.lock
+prometheus-client = "^0.24.0"
+pandas = "^2.2.0"
+numpy = "^2.2.0"
+mcp = "^1.10.0"
+pillow = "^12.1.0"
 
 [tool.poetry.group.dev.dependencies]
 ruff = "0.8.3"

enterprise/saas_server.py

@@ -78,8 +78,15 @@ base_app.include_router(shared_event_router)
 
 # Add GitHub integration router only if GITHUB_APP_CLIENT_ID is set
 if GITHUB_APP_CLIENT_ID:
+    # Make sure that the callback processor is loaded here so we don't get an error when deserializing
+    from integrations.github.github_v1_callback_processor import (  # noqa: E402
+        GithubV1CallbackProcessor,
+    )
     from server.routes.integration.github import github_integration_router  # noqa: E402
 
+    # Bludgeon mypy into not deleting my import
+    logger.debug(f'Loaded {GithubV1CallbackProcessor.__name__}')
+
     base_app.include_router(
         github_integration_router
     )  # Add additional route for integration webhook events

enterprise/server/auth/authorization.py

@@ -0,0 +1,320 @@
+"""
+Permission-based authorization dependencies for API endpoints (SAAS mode).
+
+This module provides FastAPI dependencies for checking user permissions
+within organizations. It uses a permission-based authorization model where
+roles (owner, admin, member) are mapped to specific permissions.
+
+This is the SAAS/enterprise implementation that performs real authorization
+checks against the database.
+
+Usage:
+    from server.auth.authorization import (
+        Permission,
+        require_permission,
+        require_org_role,
+        require_org_user,
+        require_org_admin,
+        require_org_owner,
+    )
+
+    @router.get('/{org_id}/settings')
+    async def get_settings(
+        org_id: UUID,
+        user_id: str = Depends(require_permission(Permission.VIEW_LLM_SETTINGS)),
+    ):
+        # Only users with VIEW_LLM_SETTINGS permission can access
+        ...
+"""
+
+from enum import Enum
+from uuid import UUID
+
+from fastapi import Depends, HTTPException, status
+from storage.org_member_store import OrgMemberStore
+from storage.role import Role
+from storage.role_store import RoleStore
+
+from openhands.core.logger import openhands_logger as logger
+from openhands.server.auth import Permission
+from openhands.server.user_auth import get_user_id
+
+
+class RoleName(str, Enum):
+    """Role names used in the system."""
+
+    OWNER = 'owner'
+    ADMIN = 'admin'
+    MEMBER = 'member'
+
+
+# Permission mappings for each role
+ROLE_PERMISSIONS: dict[RoleName, frozenset[Permission]] = {
+    RoleName.OWNER: frozenset(
+        [
+            # Settings (Full access)
+            Permission.MANAGE_SECRETS,
+            Permission.MANAGE_MCP,
+            Permission.MANAGE_INTEGRATIONS,
+            Permission.MANAGE_APPLICATION_SETTINGS,
+            Permission.MANAGE_API_KEYS,
+            Permission.VIEW_LLM_SETTINGS,
+            Permission.EDIT_LLM_SETTINGS,
+            Permission.VIEW_BILLING,
+            Permission.ADD_CREDITS,
+            # Organization Members
+            Permission.INVITE_USER_TO_ORGANIZATION,
+            Permission.CHANGE_USER_ROLE_MEMBER,
+            Permission.CHANGE_USER_ROLE_ADMIN,
+            Permission.CHANGE_USER_ROLE_OWNER,
+            # Organization Management (Owner only)
+            Permission.CHANGE_ORGANIZATION_NAME,
+            Permission.DELETE_ORGANIZATION,
+        ]
+    ),
+    RoleName.ADMIN: frozenset(
+        [
+            # Settings (Full access)
+            Permission.MANAGE_SECRETS,
+            Permission.MANAGE_MCP,
+            Permission.MANAGE_INTEGRATIONS,
+            Permission.MANAGE_APPLICATION_SETTINGS,
+            Permission.MANAGE_API_KEYS,
+            Permission.VIEW_LLM_SETTINGS,
+            Permission.EDIT_LLM_SETTINGS,
+            Permission.VIEW_BILLING,
+            Permission.ADD_CREDITS,
+            # Organization Members
+            Permission.INVITE_USER_TO_ORGANIZATION,
+            Permission.CHANGE_USER_ROLE_MEMBER,
+            Permission.CHANGE_USER_ROLE_ADMIN,
+        ]
+    ),
+    RoleName.MEMBER: frozenset(
+        [
+            # Settings (Full access)
+            Permission.MANAGE_SECRETS,
+            Permission.MANAGE_MCP,
+            Permission.MANAGE_INTEGRATIONS,
+            Permission.MANAGE_APPLICATION_SETTINGS,
+            Permission.MANAGE_API_KEYS,
+            # LLM Settings (View only)
+            Permission.VIEW_LLM_SETTINGS,
+        ]
+    ),
+}
+
+
+def get_role_permissions(role_name: str) -> frozenset[Permission]:
+    """Get the permissions for a role."""
+    try:
+        role_enum = RoleName(role_name)
+        return ROLE_PERMISSIONS.get(role_enum, frozenset())
+    except ValueError:
+        return frozenset()
+
+
+def get_user_org_role(user_id: str, org_id: UUID) -> Role | None:
+    """
+    Get the user's role in an organization.
+
+    Args:
+        user_id: User ID (string that will be converted to UUID)
+        org_id: Organization ID
+
+    Returns:
+        Role object if user is a member, None otherwise
+    """
+    from uuid import UUID as parse_uuid
+
+    org_member = OrgMemberStore.get_org_member(org_id, parse_uuid(user_id))
+    if not org_member:
+        return None
+
+    return RoleStore.get_role_by_id(org_member.role_id)
+
+
+def has_permission(user_role: Role, permission: Permission) -> bool:
+    """
+    Check if a role has a specific permission.
+
+    Args:
+        user_role: User's Role object
+        permission: Permission to check
+
+    Returns:
+        True if the role has the permission
+    """
+    permissions = get_role_permissions(user_role.name)
+    return permission in permissions
+
+
+def has_required_role(user_role: Role, required_role: Role) -> bool:
+    """
+    Check if user's role meets or exceeds the required role.
+
+    Uses role hierarchy based on rank where lower rank = higher position
+    (e.g., rank 1 owner > rank 2 admin > rank 3 user).
+
+    Args:
+        user_role: User's actual Role object
+        required_role: Minimum required Role object
+
+    Returns:
+        True if user has sufficient permissions
+    """
+    return user_role.rank <= required_role.rank
+
+
+def require_permission(permission: Permission):
+    """
+    Factory function that creates a dependency to require a specific permission.
+
+    This creates a FastAPI dependency that:
+    1. Extracts org_id from the path parameter
+    2. Gets the authenticated user_id
+    3. Checks if the user has the required permission in the organization
+    4. Returns the user_id if authorized, raises HTTPException otherwise
+
+    Usage:
+        @router.get('/{org_id}/settings')
+        async def get_settings(
+            org_id: UUID,
+            user_id: str = Depends(require_permission(Permission.VIEW_LLM_SETTINGS)),
+        ):
+            ...
+
+    Args:
+        permission: The permission required to access the endpoint
+
+    Returns:
+        Dependency function that validates permission and returns user_id
+    """
+
+    async def permission_checker(
+        org_id: UUID,
+        user_id: str | None = Depends(get_user_id),
+    ) -> str:
+        if not user_id:
+            raise HTTPException(
+                status_code=status.HTTP_401_UNAUTHORIZED,
+                detail='User not authenticated',
+            )
+
+        user_role = get_user_org_role(user_id, org_id)
+
+        if not user_role:
+            logger.warning(
+                'User not a member of organization',
+                extra={'user_id': user_id, 'org_id': str(org_id)},
+            )
+            raise HTTPException(
+                status_code=status.HTTP_403_FORBIDDEN,
+                detail='User is not a member of this organization',
+            )
+
+        if not has_permission(user_role, permission):
+            logger.warning(
+                'Insufficient permissions',
+                extra={
+                    'user_id': user_id,
+                    'org_id': str(org_id),
+                    'user_role': user_role.name,
+                    'required_permission': permission.value,
+                },
+            )
+            raise HTTPException(
+                status_code=status.HTTP_403_FORBIDDEN,
+                detail=f'Requires {permission.value} permission',
+            )
+
+        return user_id
+
+    return permission_checker
+
+
+def require_org_role(required_role_name: str):
+    """
+    Factory function that creates a dependency to require a minimum org role.
+
+    This creates a FastAPI dependency that:
+    1. Extracts org_id from the path parameter
+    2. Gets the authenticated user_id
+    3. Checks if the user has the required role in the organization
+    4. Returns the user_id if authorized, raises HTTPException otherwise
+
+    Role hierarchy is based on rank from the Role class, where
+    lower rank = higher position (e.g., rank 1 > rank 2 > rank 3).
+
+    Usage:
+        @router.get('/{org_id}/resource')
+        async def get_resource(
+            org_id: UUID,
+            user_id: str = Depends(require_org_role('user')),
+        ):
+            ...
+
+    Args:
+        required_role_name: Name of the minimum required role to access the endpoint
+
+    Returns:
+        Dependency function that validates role and returns user_id
+    """
+
+    async def role_checker(
+        org_id: UUID,
+        user_id: str | None = Depends(get_user_id),
+    ) -> str:
+        if not user_id:
+            raise HTTPException(
+                status_code=status.HTTP_401_UNAUTHORIZED,
+                detail='User not authenticated',
+            )
+
+        user_role = get_user_org_role(user_id, org_id)
+
+        if not user_role:
+            logger.warning(
+                'User not a member of organization',
+                extra={'user_id': user_id, 'org_id': str(org_id)},
+            )
+            raise HTTPException(
+                status_code=status.HTTP_403_FORBIDDEN,
+                detail='User is not a member of this organization',
+            )
+
+        required_role = RoleStore.get_role_by_name(required_role_name)
+        if not required_role:
+            logger.error(
+                'Required role not found in database',
+                extra={'required_role': required_role_name},
+            )
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail='Role configuration error',
+            )
+
+        if not has_required_role(user_role, required_role):
+            logger.warning(
+                'Insufficient role permissions',
+                extra={
+                    'user_id': user_id,
+                    'org_id': str(org_id),
+                    'user_role': user_role.name,
+                    'required_role': required_role_name,
+                },
+            )
+            raise HTTPException(
+                status_code=status.HTTP_403_FORBIDDEN,
+                detail=f'Requires {required_role_name} role or higher',
+            )
+
+        return user_id
+
+    return role_checker
+
+
+# Convenience dependencies for common role checks
+require_org_user = require_org_role('user')
+require_org_admin = require_org_role('admin')
+require_org_owner = require_org_role('owner')

enterprise/server/auth/gitlab_sync.py

@@ -1,11 +1,36 @@
 import asyncio
 
 from pydantic import SecretStr
+from sqlalchemy import select
 
 from openhands.core.logger import openhands_logger as logger
+from openhands.integrations.service_types import ProviderType
 from openhands.server.types import AppMode
 
 
+async def _user_has_gitlab_provider(user_id: str) -> bool:
+    """Check if the user has authenticated with GitLab.
+
+    Args:
+        user_id: The Keycloak user ID
+
+    Returns:
+        True if the user has a GitLab provider token, False otherwise
+    """
+    # Lazy import to avoid circular dependency issues at module load time
+    from storage.auth_tokens import AuthTokens
+    from storage.database import a_session_maker
+
+    async with a_session_maker() as session:
+        result = await session.execute(
+            select(AuthTokens).where(
+                AuthTokens.keycloak_user_id == user_id,
+                AuthTokens.identity_provider == ProviderType.GITLAB.value,
+            )
+        )
+        return result.scalars().first() is not None
+
+
 def schedule_gitlab_repo_sync(
     user_id: str, keycloak_access_token: SecretStr | None = None
 ) -> None:
@@ -14,10 +39,20 @@ def schedule_gitlab_repo_sync(
     Because the outer call is already a background task, we instruct the service
     to store repository data synchronously (store_in_background=False) to avoid
     nested background tasks while still keeping the overall operation async.
+
+    The sync is only performed if the user has authenticated with GitLab.
     """
 
     async def _run():
         try:
+            # Check if the user has a GitLab provider token before syncing
+            if not await _user_has_gitlab_provider(user_id):
+                logger.debug(
+                    'gitlab_repo_sync_skipped: user has no GitLab provider',
+                    extra={'user_id': user_id},
+                )
+                return
+
             # Lazy import to avoid circular dependency:
             # middleware -> gitlab_sync -> integrations.gitlab.gitlab_service
             # -> openhands.integrations.gitlab.gitlab_service -> get_impl

enterprise/server/auth/recaptcha_service.py

@@ -18,6 +18,7 @@ from openhands.core.logger import openhands_logger as logger
 class AssessmentResult:
     """Result of a reCAPTCHA Enterprise assessment."""
 
+    name: str
     score: float
     valid: bool
     action_valid: bool
@@ -63,6 +64,7 @@ class RecaptchaService:
         user_ip: str,
         user_agent: str,
         email: str | None = None,
+        user_id: str | None = None,
     ) -> AssessmentResult:
         """Create a reCAPTCHA Enterprise assessment.
 
@@ -72,6 +74,7 @@ class RecaptchaService:
             user_ip: The user's IP address.
             user_agent: The user's browser user agent.
             email: Optional email for Account Defender hashing.
+            user_id: Optional Keycloak user ID for logging correlation.
 
         Returns:
             AssessmentResult with score, validity, and allowed status.
@@ -100,6 +103,10 @@ class RecaptchaService:
 
         response = self.client.create_assessment(request)
 
+        # Capture assessment name for potential annotation later
+        # Format: projects/{project_id}/assessments/{assessment_id}
+        assessment_name = response.name
+
         token_properties = response.token_properties
         risk_analysis = response.risk_analysis
 
@@ -129,6 +136,7 @@ class RecaptchaService:
         logger.info(
             'recaptcha_assessment',
             extra={
+                'assessment_name': assessment_name,
                 'score': score,
                 'valid': valid,
                 'action_valid': action_valid,
@@ -137,10 +145,13 @@ class RecaptchaService:
                 'has_suspicious_labels': has_suspicious_labels,
                 'allowed': allowed,
                 'user_ip': user_ip,
+                'user_id': user_id,
+                'email': email,
             },
         )
 
         return AssessmentResult(
+            name=assessment_name,
             score=score,
             valid=valid,
             action_valid=action_valid,

enterprise/server/auth/saas_user_auth.py

@@ -216,9 +216,9 @@ class SaasUserAuth(UserAuth):
 
     async def get_mcp_api_key(self) -> str:
         api_key_store = ApiKeyStore.get_instance()
-        mcp_api_key = api_key_store.retrieve_mcp_api_key(self.user_id)
+        mcp_api_key = await api_key_store.retrieve_mcp_api_key(self.user_id)
         if not mcp_api_key:
-            mcp_api_key = api_key_store.create_api_key(
+            mcp_api_key = await api_key_store.create_api_key(
                 self.user_id, 'MCP_API_KEY', None
             )
         return mcp_api_key

enterprise/server/auth/token_manager.py

@@ -16,6 +16,7 @@ from keycloak.exceptions import (
     KeycloakError,
     KeycloakPostError,
 )
+from server.auth.auth_error import ExpiredError
 from server.auth.constants import (
     BITBUCKET_APP_CLIENT_ID,
     BITBUCKET_APP_CLIENT_SECRET,
@@ -426,6 +427,8 @@ class TokenManager:
         access_token = data.get('access_token')
         refresh_token = data.get('refresh_token')
         if not access_token or not refresh_token:
+            if data.get('error') == 'bad_refresh_token':
+                raise ExpiredError()
             raise ValueError(
                 'Failed to refresh token: missing access_token or refresh_token in response.'
             )

enterprise/server/constants.py

@@ -15,6 +15,11 @@ IS_FEATURE_ENV = (
 )  # Does not include the staging deployment
 IS_LOCAL_ENV = bool(HOST == 'localhost')
 
+# Role name constants
+ROLE_OWNER = 'owner'
+ROLE_ADMIN = 'admin'
+ROLE_MEMBER = 'member'
+
 # Deprecated - billing margins are now handled internally in litellm
 DEFAULT_BILLING_MARGIN = float(os.environ.get('DEFAULT_BILLING_MARGIN', '1.0'))
 

enterprise/server/conversation_callback_processor/github_callback_processor.py

@@ -14,7 +14,6 @@ from storage.conversation_callback import (
     ConversationCallback,
     ConversationCallbackProcessor,
 )
-from storage.database import session_maker
 
 from openhands.core.logger import openhands_logger as logger
 from openhands.core.schema.agent import AgentState
@@ -108,13 +107,10 @@ class GithubCallbackProcessor(ConversationCallbackProcessor):
                     f'[GitHub] Sent summary instruction to conversation {conversation_id} {summary_event}'
                 )
 
-                # Update the processor state
+                # Update the processor state - the outer session will commit this
                 self.send_summary_instruction = False
                 callback.set_processor(self)
                 callback.updated_at = datetime.now()
-                with session_maker() as session:
-                    session.merge(callback)
-                    session.commit()
                 return
 
             # Extract the summary from the event store
@@ -130,14 +126,15 @@ class GithubCallbackProcessor(ConversationCallbackProcessor):
 
             logger.info(f'[GitHub] Summary sent for conversation {conversation_id}')
 
-            # Mark callback as completed status
+            # Mark callback as completed status - the outer session will commit this
             callback.status = CallbackStatus.COMPLETED
             callback.updated_at = datetime.now()
-            with session_maker() as session:
-                session.merge(callback)
-                session.commit()
 
         except Exception as e:
             logger.exception(
                 f'[GitHub] Error processing conversation callback: {str(e)}'
             )
+            # Mark callback as error to prevent infinite re-invocation
+            # The outer session will commit this
+            callback.status = CallbackStatus.ERROR
+            callback.updated_at = datetime.now()

enterprise/server/mcp/mcp_config.py

@@ -22,7 +22,7 @@ from openhands.core.logger import openhands_logger as logger
 # NOTE: these details are specific to the MCP protocol
 class SaaSOpenHandsMCPConfig(OpenHandsMCPConfig):
     @staticmethod
-    def create_default_mcp_server_config(
+    async def create_default_mcp_server_config(
         host: str, config: 'OpenHandsConfig', user_id: str | None = None
     ) -> tuple[MCPSHTTPServerConfig | None, list[MCPStdioServerConfig]]:
         """
@@ -38,10 +38,12 @@ class SaaSOpenHandsMCPConfig(OpenHandsMCPConfig):
 
         api_key_store = ApiKeyStore.get_instance()
         if user_id:
-            api_key = api_key_store.retrieve_mcp_api_key(user_id)
+            api_key = await api_key_store.retrieve_mcp_api_key(user_id)
 
             if not api_key:
-                api_key = api_key_store.create_api_key(user_id, 'MCP_API_KEY', None)
+                api_key = await api_key_store.create_api_key(
+                    user_id, 'MCP_API_KEY', None
+                )
 
             if not api_key:
                 logger.error(f'Could not provision MCP API Key for user: {user_id}')

enterprise/server/middleware.py

@@ -103,7 +103,7 @@ class SetAuthCookieMiddleware:
         keycloak_auth_cookie = request.cookies.get('keycloak_auth')
         auth_header = request.headers.get('Authorization')
         mcp_auth_header = request.headers.get('X-Session-API-Key')
-        accepted_tos = False
+        accepted_tos: bool | None = False
         if (
             keycloak_auth_cookie is None
             and (auth_header is None or not auth_header.startswith('Bearer '))
@@ -144,7 +144,7 @@ class SetAuthCookieMiddleware:
         # "if accepted_tos is not None" as there should not be any users with
         # accepted_tos equal to "None"
         if accepted_tos is False and request.url.path != '/api/accept_tos':
-            logger.error('User has not accepted the terms of service')
+            logger.warning('User has not accepted the terms of service')
             raise TosNotAcceptedError
 
     def _should_attach(self, request: Request) -> bool:

enterprise/server/routes/api_keys.py

@@ -6,62 +6,53 @@ from storage.api_key_store import ApiKeyStore
 from storage.lite_llm_manager import LiteLlmManager
 from storage.org_member import OrgMember
 from storage.org_member_store import OrgMemberStore
+from storage.org_service import OrgService
 from storage.user_store import UserStore
 
 from openhands.core.logger import openhands_logger as logger
 from openhands.server.user_auth import get_user_id
-from openhands.utils.async_utils import call_sync_from_async
 
 
 # Helper functions for BYOR API key management
 async def get_byor_key_from_db(user_id: str) -> str | None:
     """Get the BYOR key from the database for a user."""
-
-    def _get_byor_key():
-        user = UserStore.get_user_by_id(user_id)
-        if not user:
-            return None
-
-        current_org_id = user.current_org_id
-        current_org_member: OrgMember = None
-        for org_member in user.org_members:
-            if org_member.org_id == current_org_id:
-                current_org_member = org_member
-                break
-        if not current_org_member:
-            return None
-        if current_org_member.llm_api_key_for_byor:
-            return current_org_member.llm_api_key_for_byor.get_secret_value()
+    user = await UserStore.get_user_by_id_async(user_id)
+    if not user:
         return None
 
-    return await call_sync_from_async(_get_byor_key)
+    current_org_id = user.current_org_id
+    current_org_member: OrgMember = None
+    for org_member in user.org_members:
+        if org_member.org_id == current_org_id:
+            current_org_member = org_member
+            break
+    if not current_org_member:
+        return None
+    if current_org_member.llm_api_key_for_byor:
+        return current_org_member.llm_api_key_for_byor.get_secret_value()
+    return None
 
 
 async def store_byor_key_in_db(user_id: str, key: str) -> None:
     """Store the BYOR key in the database for a user."""
+    user = await UserStore.get_user_by_id_async(user_id)
+    if not user:
+        return None
 
-    def _update_user_settings():
-        user = UserStore.get_user_by_id(user_id)
-        if not user:
-            return None
-
-        current_org_id = user.current_org_id
-        current_org_member: OrgMember = None
-        for org_member in user.org_members:
-            if org_member.org_id == current_org_id:
-                current_org_member = org_member
-                break
-        if not current_org_member:
-            return None
-        current_org_member.llm_api_key_for_byor = key
-        OrgMemberStore.update_org_member(current_org_member)
-
-    await call_sync_from_async(_update_user_settings)
+    current_org_id = user.current_org_id
+    current_org_member: OrgMember = None
+    for org_member in user.org_members:
+        if org_member.org_id == current_org_id:
+            current_org_member = org_member
+            break
+    if not current_org_member:
+        return None
+    current_org_member.llm_api_key_for_byor = key
+    OrgMemberStore.update_org_member(current_org_member)
 
 
 async def generate_byor_key(user_id: str) -> str | None:
     """Generate a new BYOR key for a user."""
-
     try:
         user = await UserStore.get_user_by_id_async(user_id)
         if not user:
@@ -157,15 +148,35 @@ class LlmApiKeyResponse(BaseModel):
     key: str | None
 
 
+class ByorPermittedResponse(BaseModel):
+    permitted: bool
+
+
+@api_router.get('/llm/byor/permitted', response_model=ByorPermittedResponse)
+async def check_byor_permitted(user_id: str = Depends(get_user_id)):
+    """Check if BYOR key export is permitted for the user's current org."""
+    try:
+        permitted = await OrgService.check_byor_export_enabled(user_id)
+        return {'permitted': permitted}
+    except Exception as e:
+        logger.exception(
+            'Error checking BYOR export permission', extra={'error': str(e)}
+        )
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='Failed to check BYOR export permission',
+        )
+
+
 @api_router.post('', response_model=ApiKeyCreateResponse)
 async def create_api_key(key_data: ApiKeyCreate, user_id: str = Depends(get_user_id)):
     """Create a new API key for the authenticated user."""
     try:
-        api_key = api_key_store.create_api_key(
+        api_key = await api_key_store.create_api_key(
             user_id, key_data.name, key_data.expires_at
         )
         # Get the created key details
-        keys = api_key_store.list_api_keys(user_id)
+        keys = await api_key_store.list_api_keys(user_id)
         for key in keys:
             if key['name'] == key_data.name:
                 return {
@@ -193,7 +204,7 @@ async def create_api_key(key_data: ApiKeyCreate, user_id: str = Depends(get_user
 async def list_api_keys(user_id: str = Depends(get_user_id)):
     """List all API keys for the authenticated user."""
     try:
-        keys = api_key_store.list_api_keys(user_id)
+        keys = await api_key_store.list_api_keys(user_id)
         return [
             {
                 **key,
@@ -222,7 +233,7 @@ async def delete_api_key(key_id: int, user_id: str = Depends(get_user_id)):
     """Delete an API key."""
     try:
         # First, verify the key belongs to the user
-        keys = api_key_store.list_api_keys(user_id)
+        keys = await api_key_store.list_api_keys(user_id)
         key_to_delete = None
 
         for key in keys:
@@ -262,8 +273,17 @@ async def get_llm_api_key_for_byor(user_id: str = Depends(get_user_id)):
     This endpoint validates that the key exists in LiteLLM before returning it.
     If validation fails, it automatically generates a new key to ensure users
     always receive a working key.
+
+    Returns 402 Payment Required if BYOR export is not enabled for the user's org.
     """
     try:
+        # Check if BYOR export is enabled for the user's org
+        if not await OrgService.check_byor_export_enabled(user_id):
+            raise HTTPException(
+                status_code=status.HTTP_402_PAYMENT_REQUIRED,
+                detail='BYOR key export is not enabled. Purchase credits to enable this feature.',
+            )
+
         # Check if the BYOR key exists in the database
         byor_key = await get_byor_key_from_db(user_id)
         if byor_key:
@@ -319,10 +339,20 @@ async def get_llm_api_key_for_byor(user_id: str = Depends(get_user_id)):
 
 @api_router.post('/llm/byor/refresh', response_model=LlmApiKeyResponse)
 async def refresh_llm_api_key_for_byor(user_id: str = Depends(get_user_id)):
-    """Refresh the LLM API key for BYOR (Bring Your Own Runtime) for the authenticated user."""
+    """Refresh the LLM API key for BYOR (Bring Your Own Runtime) for the authenticated user.
+
+    Returns 402 Payment Required if BYOR export is not enabled for the user's org.
+    """
     logger.info('Starting BYOR LLM API key refresh', extra={'user_id': user_id})
 
     try:
+        # Check if BYOR export is enabled for the user's org
+        if not await OrgService.check_byor_export_enabled(user_id):
+            raise HTTPException(
+                status_code=status.HTTP_402_PAYMENT_REQUIRED,
+                detail='BYOR key export is not enabled. Purchase credits to enable this feature.',
+            )
+
         # Get the existing BYOR key from the database
         existing_byor_key = await get_byor_key_from_db(user_id)
 

enterprise/server/routes/auth.py

@@ -179,6 +179,9 @@ async def keycloak_callback(
     user = await UserStore.get_user_by_id_async(user_id)
     if not user:
         user = await UserStore.create_user(user_id, user_info)
+    else:
+        # Existing user — gradually backfill contact_name if it still has a username-style value
+        await UserStore.backfill_contact_name(user_id, user_info)
 
     if not user:
         logger.error(f'Failed to authenticate user {user_info["preferred_username"]}')
@@ -219,6 +222,7 @@ async def keycloak_callback(
                 user_ip=user_ip,
                 user_agent=user_agent,
                 email=email,
+                user_id=user_id,
             )
 
             if not result.allowed:

enterprise/server/routes/billing.py

@@ -13,46 +13,34 @@ from server.constants import (
     STRIPE_API_KEY,
 )
 from server.logger import logger
+from starlette.datastructures import URL
 from storage.billing_session import BillingSession
 from storage.database import session_maker
 from storage.lite_llm_manager import LiteLlmManager
+from storage.org import Org
 from storage.subscription_access import SubscriptionAccess
 from storage.user_store import UserStore
 
+from openhands.app_server.config import get_global_config
 from openhands.server.user_auth import get_user_id
 
 stripe.api_key = STRIPE_API_KEY
 billing_router = APIRouter(prefix='/api/billing')
 
 
-# TODO: Add a new app_mode named "ON_PREM" to support self-hosted customers instead of doing this
-# and members should comment out the "validate_saas_environment" function if they are developing and testing locally.
-def is_all_hands_saas_environment(request: Request) -> bool:
-    """Check if the current domain is an All Hands SaaS environment.
-
-    Args:
-        request: FastAPI Request object
-
-    Returns:
-        True if the current domain contains "all-hands.dev" or "openhands.dev" postfix
+async def validate_billing_enabled() -> None:
     """
-    hostname = request.url.hostname or ''
-    return hostname.endswith('all-hands.dev') or hostname.endswith('openhands.dev')
-
-
-def validate_saas_environment(request: Request) -> None:
-    """Validate that the request is coming from an All Hands SaaS environment.
-
-    Args:
-        request: FastAPI Request object
-
-    Raises:
-        HTTPException: If the request is not from an All Hands SaaS environment
+    Validate that the billing feature flag is enabled
     """
-    if not is_all_hands_saas_environment(request):
+    config = get_global_config()
+    web_client_config = await config.web_client.get_web_client_config()
+    if not web_client_config.feature_flags.enable_billing:
         raise HTTPException(
             status_code=status.HTTP_403_FORBIDDEN,
-            detail='Checkout sessions are only available for All Hands SaaS environments',
+            detail=(
+                'Billing is disabled in this environment. '
+                'Please set OH_WEB_CLIENT_FEATURE_FLAGS_ENABLE_BILLING to enable billing.'
+            ),
         )
 
 
@@ -154,14 +142,15 @@ async def has_payment_method(user_id: str = Depends(get_user_id)) -> bool:
 async def create_customer_setup_session(
     request: Request, user_id: str = Depends(get_user_id)
 ) -> CreateBillingSessionResponse:
-    validate_saas_environment(request)
+    await validate_billing_enabled()
     customer_info = await stripe_service.find_or_create_customer_by_user_id(user_id)
+    base_url = _get_base_url(request)
     checkout_session = await stripe.checkout.Session.create_async(
         customer=customer_info['customer_id'],
         mode='setup',
         payment_method_types=['card'],
-        success_url=f'{request.base_url}?free_credits=success',
-        cancel_url=f'{request.base_url}',
+        success_url=f'{base_url}?free_credits=success',
+        cancel_url=f'{base_url}',
     )
     return CreateBillingSessionResponse(redirect_url=checkout_session.url)
 
@@ -173,8 +162,8 @@ async def create_checkout_session(
     request: Request,
     user_id: str = Depends(get_user_id),
 ) -> CreateBillingSessionResponse:
-    validate_saas_environment(request)
-
+    await validate_billing_enabled()
+    base_url = _get_base_url(request)
     customer_info = await stripe_service.find_or_create_customer_by_user_id(user_id)
     checkout_session = await stripe.checkout.Session.create_async(
         customer=customer_info['customer_id'],
@@ -197,8 +186,8 @@ async def create_checkout_session(
         saved_payment_method_options={
             'payment_method_save': 'enabled',
         },
-        success_url=f'{request.base_url}api/billing/success?session_id={{CHECKOUT_SESSION_ID}}',
-        cancel_url=f'{request.base_url}api/billing/cancel?session_id={{CHECKOUT_SESSION_ID}}',
+        success_url=f'{base_url}api/billing/success?session_id={{CHECKOUT_SESSION_ID}}',
+        cancel_url=f'{base_url}api/billing/cancel?session_id={{CHECKOUT_SESSION_ID}}',
     )
     logger.info(
         'created_stripe_checkout_session',
@@ -271,6 +260,12 @@ async def success_callback(session_id: str, request: Request):
             str(user.current_org_id), new_max_budget
         )
 
+        # Enable BYOR export for the org now that they've purchased credits
+        # Update within the same session to avoid nested session issues
+        org = session.query(Org).filter(Org.id == user.current_org_id).first()
+        if org:
+            org.byor_export_enabled = True
+
         # Store transaction status
         billing_session.status = 'completed'
         billing_session.price = add_credits
@@ -289,7 +284,7 @@ async def success_callback(session_id: str, request: Request):
         session.commit()
 
     return RedirectResponse(
-        f'{request.base_url}settings/billing?checkout=success', status_code=302
+        f'{_get_base_url(request)}settings/billing?checkout=success', status_code=302
     )
 
 
@@ -317,5 +312,13 @@ async def cancel_callback(session_id: str, request: Request):
             session.commit()
 
     return RedirectResponse(
-        f'{request.base_url}settings/billing?checkout=cancel', status_code=302
+        f'{_get_base_url(request)}settings/billing?checkout=cancel', status_code=302
     )
+
+
+def _get_base_url(request: Request) -> URL:
+    # Never send any part of the credit card process over a non secure connection
+    base_url = request.base_url
+    if base_url.hostname != 'localhost':
+        base_url = base_url.replace(scheme='https')
+    return base_url

enterprise/server/routes/github_proxy.py

@@ -1,6 +1,7 @@
 import hashlib
 import json
 import os
+import zlib
 from base64 import b64decode, b64encode
 from urllib.parse import parse_qs, urlencode, urlparse
 
@@ -51,7 +52,11 @@ def add_github_proxy_routes(app: FastAPI):
         state_payload = json.dumps(
             [query_params['state'][0], query_params['redirect_uri'][0]]
         )
-        state = b64encode(_fernet().encrypt(state_payload.encode())).decode()
+        # Compress before encrypting to reduce URL length
+        # This is critical for feature deployments where reCAPTCHA tokens in state
+        # can cause "URL too long" errors from GitHub
+        compressed_payload = zlib.compress(state_payload.encode())
+        state = b64encode(_fernet().encrypt(compressed_payload)).decode()
         query_params['state'] = [state]
         query_params['redirect_uri'] = [
             f'https://{request.url.netloc}/github-proxy/callback'
@@ -67,7 +72,9 @@ def add_github_proxy_routes(app: FastAPI):
         parsed_url = urlparse(str(request.url))
         query_params = parse_qs(parsed_url.query)
         state = query_params['state'][0]
-        decrypted_state = _fernet().decrypt(b64decode(state.encode())).decode()
+        # Decrypt and decompress (reverse of github_proxy_start)
+        decrypted_payload = _fernet().decrypt(b64decode(state.encode()))
+        decrypted_state = zlib.decompress(decrypted_payload).decode()
 
         # Build query Params
         state, redirect_uri = json.loads(decrypted_state)

enterprise/server/routes/oauth_device.py

@@ -272,7 +272,7 @@ async def device_verification_authenticated(
         try:
             # Create a unique API key for this device using user_code in the name
             device_key_name = f'{API_KEY_NAME} ({user_code})'
-            api_key_store.create_api_key(
+            await api_key_store.create_api_key(
                 user_id,
                 name=device_key_name,
                 expires_at=datetime.now(UTC) + KEY_EXPIRATION_TIME,

enterprise/server/routes/org_models.py

@@ -1,4 +1,9 @@
-from pydantic import BaseModel, EmailStr, Field
+from typing import Annotated
+
+from pydantic import BaseModel, EmailStr, Field, SecretStr, StringConstraints
+from storage.org import Org
+from storage.org_member import OrgMember
+from storage.role import Role
 
 
 class OrgCreationError(Exception):
@@ -27,13 +32,101 @@ class OrgDatabaseError(OrgCreationError):
     pass
 
 
+class OrgDeletionError(Exception):
+    """Base exception for organization deletion errors."""
+
+    pass
+
+
+class OrgAuthorizationError(OrgDeletionError):
+    """Raised when user is not authorized to delete organization."""
+
+    def __init__(self, message: str = 'Not authorized to delete organization'):
+        super().__init__(message)
+
+
+class OrphanedUserError(OrgDeletionError):
+    """Raised when deleting an org would leave users without any organization."""
+
+    def __init__(self, user_ids: list[str]):
+        self.user_ids = user_ids
+        super().__init__(
+            f'Cannot delete organization: {len(user_ids)} user(s) would have no remaining organization'
+        )
+
+
+class OrgNotFoundError(Exception):
+    """Raised when organization is not found or user doesn't have access."""
+
+    def __init__(self, org_id: str):
+        self.org_id = org_id
+        super().__init__(f'Organization with id "{org_id}" not found')
+
+
+class OrgMemberNotFoundError(Exception):
+    """Raised when a member is not found in an organization."""
+
+    def __init__(self, org_id: str, user_id: str):
+        self.org_id = org_id
+        self.user_id = user_id
+        super().__init__(f'Member "{user_id}" not found in organization "{org_id}"')
+
+
+class RoleNotFoundError(Exception):
+    """Raised when a role is not found."""
+
+    def __init__(self, role_id: int):
+        self.role_id = role_id
+        super().__init__(f'Role with id "{role_id}" not found')
+
+
+class InvalidRoleError(Exception):
+    """Raised when an invalid role name is specified."""
+
+    def __init__(self, role_name: str):
+        self.role_name = role_name
+        super().__init__(f'Invalid role: "{role_name}"')
+
+
+class InsufficientPermissionError(Exception):
+    """Raised when user lacks permission to perform an operation."""
+
+    def __init__(self, message: str = 'Insufficient permission'):
+        super().__init__(message)
+
+
+class CannotModifySelfError(Exception):
+    """Raised when user attempts to modify their own membership."""
+
+    def __init__(self, action: str = 'modify'):
+        self.action = action
+        super().__init__(f'Cannot {action} your own membership')
+
+
+class LastOwnerError(Exception):
+    """Raised when attempting to remove or demote the last owner."""
+
+    def __init__(self, action: str = 'remove'):
+        self.action = action
+        super().__init__(f'Cannot {action} the last owner of an organization')
+
+
+class MemberUpdateError(Exception):
+    """Raised when member update operation fails."""
+
+    def __init__(self, message: str = 'Failed to update member'):
+        super().__init__(message)
+
+
 class OrgCreate(BaseModel):
     """Request model for creating a new organization."""
 
     # Required fields
-    name: str = Field(min_length=1, max_length=255, strip_whitespace=True)
+    name: Annotated[
+        str, StringConstraints(strip_whitespace=True, min_length=1, max_length=255)
+    ]
     contact_name: str
-    contact_email: EmailStr = Field(strip_whitespace=True)
+    contact_email: EmailStr
 
 
 class OrgResponse(BaseModel):
@@ -65,3 +158,170 @@ class OrgResponse(BaseModel):
     enable_solvability_analysis: bool | None = None
     v1_enabled: bool | None = None
     credits: float | None = None
+    is_personal: bool = False
+
+    @classmethod
+    def from_org(
+        cls, org: Org, credits: float | None = None, user_id: str | None = None
+    ) -> 'OrgResponse':
+        """Create an OrgResponse from an Org entity.
+
+        Args:
+            org: The organization entity to convert
+            credits: Optional credits value (defaults to None)
+            user_id: Optional user ID to determine if org is personal (defaults to None)
+
+        Returns:
+            OrgResponse: The response model instance
+        """
+        return cls(
+            id=str(org.id),
+            name=org.name,
+            contact_name=org.contact_name,
+            contact_email=org.contact_email,
+            conversation_expiration=org.conversation_expiration,
+            agent=org.agent,
+            default_max_iterations=org.default_max_iterations,
+            security_analyzer=org.security_analyzer,
+            confirmation_mode=org.confirmation_mode,
+            default_llm_model=org.default_llm_model,
+            default_llm_api_key_for_byor=None,
+            default_llm_base_url=org.default_llm_base_url,
+            remote_runtime_resource_factor=org.remote_runtime_resource_factor,
+            enable_default_condenser=org.enable_default_condenser
+            if org.enable_default_condenser is not None
+            else True,
+            billing_margin=org.billing_margin,
+            enable_proactive_conversation_starters=org.enable_proactive_conversation_starters
+            if org.enable_proactive_conversation_starters is not None
+            else True,
+            sandbox_base_container_image=org.sandbox_base_container_image,
+            sandbox_runtime_container_image=org.sandbox_runtime_container_image,
+            org_version=org.org_version if org.org_version is not None else 0,
+            mcp_config=org.mcp_config,
+            search_api_key=None,
+            sandbox_api_key=None,
+            max_budget_per_task=org.max_budget_per_task,
+            enable_solvability_analysis=org.enable_solvability_analysis,
+            v1_enabled=org.v1_enabled,
+            credits=credits,
+            is_personal=str(org.id) == user_id if user_id else False,
+        )
+
+
+class OrgPage(BaseModel):
+    """Paginated response model for organization list."""
+
+    items: list[OrgResponse]
+    next_page_id: str | None = None
+
+
+class OrgUpdate(BaseModel):
+    """Request model for updating an organization."""
+
+    # Basic organization information (any authenticated user can update)
+    name: Annotated[
+        str | None,
+        StringConstraints(strip_whitespace=True, min_length=1, max_length=255),
+    ] = None
+    contact_name: str | None = None
+    contact_email: EmailStr | None = None
+    conversation_expiration: int | None = None
+    default_max_iterations: int | None = Field(default=None, gt=0)
+    remote_runtime_resource_factor: int | None = Field(default=None, gt=0)
+    billing_margin: float | None = Field(default=None, ge=0, le=1)
+    enable_proactive_conversation_starters: bool | None = None
+    sandbox_base_container_image: str | None = None
+    sandbox_runtime_container_image: str | None = None
+    mcp_config: dict | None = None
+    sandbox_api_key: str | None = None
+    max_budget_per_task: float | None = Field(default=None, gt=0)
+    enable_solvability_analysis: bool | None = None
+    v1_enabled: bool | None = None
+
+    # LLM settings (require admin/owner role)
+    default_llm_model: str | None = None
+    default_llm_api_key_for_byor: str | None = None
+    default_llm_base_url: str | None = None
+    search_api_key: str | None = None
+    security_analyzer: str | None = None
+    agent: str | None = None
+    confirmation_mode: bool | None = None
+    enable_default_condenser: bool | None = None
+    condenser_max_size: int | None = Field(default=None, ge=20)
+
+
+class OrgMemberResponse(BaseModel):
+    """Response model for a single organization member."""
+
+    user_id: str
+    email: str | None
+    role_id: int
+    role_name: str
+    role_rank: int
+    status: str | None
+
+
+class OrgMemberPage(BaseModel):
+    """Paginated response for organization members."""
+
+    items: list[OrgMemberResponse]
+    next_page_id: str | None = None
+
+
+class OrgMemberUpdate(BaseModel):
+    """Request model for updating an organization member."""
+
+    role: str | None = None  # Role name: 'owner', 'admin', or 'member'
+
+
+class MeResponse(BaseModel):
+    """Response model for the current user's membership in an organization."""
+
+    org_id: str
+    user_id: str
+    email: str
+    role: str
+    llm_api_key: str
+    max_iterations: int | None = None
+    llm_model: str | None = None
+    llm_api_key_for_byor: str | None = None
+    llm_base_url: str | None = None
+    status: str | None = None
+
+    @staticmethod
+    def _mask_key(secret: SecretStr | None) -> str:
+        """Mask an API key, showing only last 4 characters."""
+        if secret is None:
+            return ''
+        raw = secret.get_secret_value()
+        if not raw:
+            return ''
+        if len(raw) <= 4:
+            return '****'
+        return '****' + raw[-4:]
+
+    @classmethod
+    def from_org_member(cls, member: OrgMember, role: Role, email: str) -> 'MeResponse':
+        """Create a MeResponse from an OrgMember, Role, and user email.
+
+        Args:
+            member: The OrgMember entity
+            role: The Role entity (provides role name)
+            email: The user's email address
+
+        Returns:
+            MeResponse with masked API keys
+        """
+        return cls(
+            org_id=str(member.org_id),
+            user_id=str(member.user_id),
+            email=email,
+            role=role.name,
+            llm_api_key=cls._mask_key(member.llm_api_key),
+            max_iterations=member.max_iterations,
+            llm_model=member.llm_model,
+            llm_api_key_for_byor=cls._mask_key(member.llm_api_key_for_byor) or None,
+            llm_base_url=member.llm_base_url,
+            status=member.status,
+        )

enterprise/server/routes/orgs.py

@@ -1,20 +1,118 @@
-from fastapi import APIRouter, Depends, HTTPException, status
+from typing import Annotated
+from uuid import UUID
+
+from fastapi import APIRouter, Depends, HTTPException, Query, status
+from server.auth.authorization import (
+    require_org_admin,
+    require_org_owner,
+    require_org_user,
+)
 from server.email_validation import get_admin_user_id
 from server.routes.org_models import (
+    CannotModifySelfError,
+    InsufficientPermissionError,
+    InvalidRoleError,
+    LastOwnerError,
     LiteLLMIntegrationError,
+    MemberUpdateError,
+    MeResponse,
+    OrgAuthorizationError,
     OrgCreate,
     OrgDatabaseError,
+    OrgMemberNotFoundError,
+    OrgMemberPage,
+    OrgMemberResponse,
+    OrgMemberUpdate,
     OrgNameExistsError,
+    OrgNotFoundError,
+    OrgPage,
     OrgResponse,
+    OrgUpdate,
+    OrphanedUserError,
+    RoleNotFoundError,
 )
+from server.services.org_member_service import OrgMemberService
 from storage.org_service import OrgService
 
 from openhands.core.logger import openhands_logger as logger
+from openhands.server.user_auth import get_user_id
 
 # Initialize API router
 org_router = APIRouter(prefix='/api/organizations')
 
 
+@org_router.get('', response_model=OrgPage)
+async def list_user_orgs(
+    page_id: Annotated[
+        str | None,
+        Query(title='Optional next_page_id from the previously returned page'),
+    ] = None,
+    limit: Annotated[
+        int,
+        Query(title='The max number of results in the page', gt=0, lte=100),
+    ] = 100,
+    user_id: str = Depends(get_user_id),
+) -> OrgPage:
+    """List organizations for the authenticated user.
+
+    This endpoint returns a paginated list of all organizations that the
+    authenticated user is a member of.
+
+    Args:
+        page_id: Optional page ID (offset) for pagination
+        limit: Maximum number of organizations to return (1-100, default 100)
+        user_id: Authenticated user ID (injected by dependency)
+
+    Returns:
+        OrgPage: Paginated list of organizations
+
+    Raises:
+        HTTPException: 500 if retrieval fails
+    """
+    logger.info(
+        'Listing organizations for user',
+        extra={
+            'user_id': user_id,
+            'page_id': page_id,
+            'limit': limit,
+        },
+    )
+
+    try:
+        # Fetch organizations from service layer
+        orgs, next_page_id = OrgService.get_user_orgs_paginated(
+            user_id=user_id,
+            page_id=page_id,
+            limit=limit,
+        )
+
+        # Convert Org entities to OrgResponse objects
+        org_responses = [
+            OrgResponse.from_org(org, credits=None, user_id=user_id) for org in orgs
+        ]
+
+        logger.info(
+            'Successfully retrieved organizations',
+            extra={
+                'user_id': user_id,
+                'org_count': len(org_responses),
+                'has_more': next_page_id is not None,
+            },
+        )
+
+        return OrgPage(items=org_responses, next_page_id=next_page_id)
+
+    except Exception as e:
+        logger.exception(
+            'Unexpected error listing organizations',
+            extra={'user_id': user_id, 'error': str(e)},
+        )
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='Failed to retrieve organizations',
+        )
+
+
 @org_router.post('', response_model=OrgResponse, status_code=status.HTTP_201_CREATED)
 async def create_org(
     org_data: OrgCreate,
@@ -58,31 +156,7 @@ async def create_org(
         # Retrieve credits from LiteLLM
         credits = await OrgService.get_org_credits(user_id, org.id)
 
-        return OrgResponse(
-            id=str(org.id),
-            name=org.name,
-            contact_name=org.contact_name,
-            contact_email=org.contact_email,
-            conversation_expiration=org.conversation_expiration,
-            agent=org.agent,
-            default_max_iterations=org.default_max_iterations,
-            security_analyzer=org.security_analyzer,
-            confirmation_mode=org.confirmation_mode,
-            default_llm_model=org.default_llm_model,
-            default_llm_base_url=org.default_llm_base_url,
-            remote_runtime_resource_factor=org.remote_runtime_resource_factor,
-            enable_default_condenser=org.enable_default_condenser,
-            billing_margin=org.billing_margin,
-            enable_proactive_conversation_starters=org.enable_proactive_conversation_starters,
-            sandbox_base_container_image=org.sandbox_base_container_image,
-            sandbox_runtime_container_image=org.sandbox_runtime_container_image,
-            org_version=org.org_version,
-            mcp_config=org.mcp_config,
-            max_budget_per_task=org.max_budget_per_task,
-            enable_solvability_analysis=org.enable_solvability_analysis,
-            v1_enabled=org.v1_enabled,
-            credits=credits,
-        )
+        return OrgResponse.from_org(org, credits=credits, user_id=user_id)
     except OrgNameExistsError as e:
         raise HTTPException(
             status_code=status.HTTP_409_CONFLICT,
@@ -115,3 +189,607 @@ async def create_org(
             status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
             detail='An unexpected error occurred',
         )
+
+
+@org_router.get('/{org_id}', response_model=OrgResponse, status_code=status.HTTP_200_OK)
+async def get_org(
+    org_id: UUID,
+    user_id: str = Depends(require_org_user),
+) -> OrgResponse:
+    """Get organization details by ID.
+
+    This endpoint allows authenticated users who are members of an organization
+    to retrieve its details. Only members of the organization can access this endpoint.
+    Requires user, admin, or owner role.
+
+    Args:
+        org_id: Organization ID (UUID)
+        user_id: Authenticated user ID (injected by dependency, requires org membership)
+
+    Returns:
+        OrgResponse: The organization details
+
+    Raises:
+        HTTPException: 401 if user is not authenticated
+        HTTPException: 403 if user is not a member of the organization
+        HTTPException: 422 if org_id is not a valid UUID (handled by FastAPI)
+        HTTPException: 404 if organization not found
+        HTTPException: 500 if retrieval fails
+    """
+    logger.info(
+        'Retrieving organization details',
+        extra={
+            'user_id': user_id,
+            'org_id': str(org_id),
+        },
+    )
+
+    try:
+        # Use service layer to get organization with membership validation
+        org = await OrgService.get_org_by_id(
+            org_id=org_id,
+            user_id=user_id,
+        )
+
+        # Retrieve credits from LiteLLM
+        credits = await OrgService.get_org_credits(user_id, org.id)
+
+        return OrgResponse.from_org(org, credits=credits, user_id=user_id)
+    except OrgNotFoundError as e:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=str(e),
+        )
+    except Exception as e:
+        logger.exception(
+            'Unexpected error retrieving organization',
+            extra={'user_id': user_id, 'org_id': str(org_id), 'error': str(e)},
+        )
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='An unexpected error occurred',
+        )
+
+
+@org_router.get('/{org_id}/me', response_model=MeResponse)
+async def get_me(
+    org_id: UUID,
+    user_id: str = Depends(get_user_id),
+) -> MeResponse:
+    """Get the current user's membership record for an organization.
+
+    Returns the authenticated user's role, status, email, and LLM override
+    fields (with masked API keys) within the specified organization.
+
+    Args:
+        org_id: Organization ID (UUID)
+        user_id: Authenticated user ID (injected by dependency)
+
+    Returns:
+        MeResponse: The user's membership data
+
+    Raises:
+        HTTPException: 404 if user is not a member or org doesn't exist
+        HTTPException: 500 if retrieval fails
+    """
+    logger.info(
+        'Retrieving current member details',
+        extra={'user_id': user_id, 'org_id': str(org_id)},
+    )
+
+    try:
+        user_uuid = UUID(user_id)
+        return OrgMemberService.get_me(org_id, user_uuid)
+
+    except OrgMemberNotFoundError:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f'Organization with id "{org_id}" not found',
+        )
+    except RoleNotFoundError as e:
+        logger.exception(
+            'Role not found for org member',
+            extra={
+                'user_id': user_id,
+                'org_id': str(org_id),
+                'role_id': e.role_id,
+            },
+        )
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='An unexpected error occurred',
+        )
+    except Exception as e:
+        logger.exception(
+            'Unexpected error retrieving member details',
+            extra={'user_id': user_id, 'org_id': str(org_id), 'error': str(e)},
+        )
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='An unexpected error occurred',
+        )
+
+
+@org_router.delete('/{org_id}', status_code=status.HTTP_200_OK)
+async def delete_org(
+    org_id: UUID,
+    user_id: str = Depends(require_org_owner),
+) -> dict:
+    """Delete an organization.
+
+    This endpoint allows authenticated organization owners to delete their organization.
+    All associated data including organization members, conversations, billing data,
+    and external LiteLLM team resources will be permanently removed.
+    Requires owner role.
+
+    Args:
+        org_id: Organization ID to delete
+        user_id: Authenticated user ID (injected by dependency, requires owner role)
+
+    Returns:
+        dict: Confirmation message with deleted organization details
+
+    Raises:
+        HTTPException: 401 if user is not authenticated
+        HTTPException: 403 if user is not an owner of the organization
+        HTTPException: 404 if organization not found
+        HTTPException: 500 if deletion fails
+    """
+    logger.info(
+        'Organization deletion requested',
+        extra={
+            'user_id': user_id,
+            'org_id': str(org_id),
+        },
+    )
+
+    try:
+        # Use service layer to delete organization with cleanup
+        deleted_org = await OrgService.delete_org_with_cleanup(
+            user_id=user_id,
+            org_id=org_id,
+        )
+
+        logger.info(
+            'Organization deletion completed successfully',
+            extra={
+                'user_id': user_id,
+                'org_id': str(org_id),
+                'org_name': deleted_org.name,
+            },
+        )
+
+        return {
+            'message': 'Organization deleted successfully',
+            'organization': {
+                'id': str(deleted_org.id),
+                'name': deleted_org.name,
+                'contact_name': deleted_org.contact_name,
+                'contact_email': deleted_org.contact_email,
+            },
+        }
+
+    except OrgNotFoundError as e:
+        logger.warning(
+            'Organization not found for deletion',
+            extra={'user_id': user_id, 'org_id': str(org_id)},
+        )
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=str(e),
+        )
+    except OrgAuthorizationError as e:
+        logger.warning(
+            'User not authorized to delete organization',
+            extra={'user_id': user_id, 'org_id': str(org_id)},
+        )
+        raise HTTPException(
+            status_code=status.HTTP_403_FORBIDDEN,
+            detail=str(e),
+        )
+    except OrphanedUserError as e:
+        logger.warning(
+            'Cannot delete organization: users would be orphaned',
+            extra={
+                'user_id': user_id,
+                'org_id': str(org_id),
+                'orphaned_users': e.user_ids,
+            },
+        )
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=str(e),
+        )
+    except OrgDatabaseError as e:
+        logger.error(
+            'Database error during organization deletion',
+            extra={'user_id': user_id, 'org_id': str(org_id), 'error': str(e)},
+        )
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='Failed to delete organization',
+        )
+    except Exception as e:
+        logger.exception(
+            'Unexpected error during organization deletion',
+            extra={'user_id': user_id, 'org_id': str(org_id), 'error': str(e)},
+        )
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='An unexpected error occurred',
+        )
+
+
+@org_router.patch('/{org_id}', response_model=OrgResponse)
+async def update_org(
+    org_id: UUID,
+    update_data: OrgUpdate,
+    user_id: str = Depends(require_org_admin),
+) -> OrgResponse:
+    """Update an existing organization.
+
+    This endpoint allows authenticated admins and owners to update organization settings.
+    Requires admin or owner role in the organization.
+
+    Args:
+        org_id: Organization ID to update (UUID validated by FastAPI)
+        update_data: Organization update data
+        user_id: Authenticated user ID (injected by dependency, requires admin role)
+
+    Returns:
+        OrgResponse: The updated organization details
+
+    Raises:
+        HTTPException: 401 if user is not authenticated
+        HTTPException: 403 if user is not an admin or owner of the organization
+        HTTPException: 404 if organization not found
+        HTTPException: 422 if validation errors occur (handled by FastAPI)
+        HTTPException: 500 if update fails
+    """
+    logger.info(
+        'Updating organization',
+        extra={
+            'user_id': user_id,
+            'org_id': str(org_id),
+        },
+    )
+
+    try:
+        # Use service layer to update organization with permission checks
+        updated_org = await OrgService.update_org_with_permissions(
+            org_id=org_id,
+            update_data=update_data,
+            user_id=user_id,
+        )
+
+        # Retrieve credits from LiteLLM (following same pattern as create endpoint)
+        credits = await OrgService.get_org_credits(user_id, updated_org.id)
+
+        return OrgResponse.from_org(updated_org, credits=credits, user_id=user_id)
+
+    except ValueError as e:
+        # Organization not found
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=str(e),
+        )
+    except OrgNameExistsError as e:
+        raise HTTPException(
+            status_code=status.HTTP_409_CONFLICT,
+            detail=str(e),
+        )
+    except PermissionError as e:
+        # User lacks permission for LLM settings
+        raise HTTPException(
+            status_code=status.HTTP_403_FORBIDDEN,
+            detail=str(e),
+        )
+    except OrgDatabaseError as e:
+        logger.error(
+            'Database operation failed',
+            extra={'user_id': user_id, 'org_id': str(org_id), 'error': str(e)},
+        )
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='Failed to update organization',
+        )
+    except Exception as e:
+        logger.exception(
+            'Unexpected error updating organization',
+            extra={'user_id': user_id, 'org_id': str(org_id), 'error': str(e)},
+        )
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='An unexpected error occurred',
+        )
+
+
+@org_router.get('/{org_id}/members')
+async def get_org_members(
+    org_id: str,
+    page_id: Annotated[
+        str | None,
+        Query(title='Optional next_page_id from the previously returned page'),
+    ] = None,
+    limit: Annotated[
+        int,
+        Query(
+            title='The max number of results in the page',
+            gt=0,
+            lte=100,
+        ),
+    ] = 100,
+    current_user_id: str = Depends(get_user_id),
+) -> OrgMemberPage:
+    """Get all members of an organization with cursor-based pagination."""
+    try:
+        success, error_code, data = await OrgMemberService.get_org_members(
+            org_id=UUID(org_id),
+            current_user_id=UUID(current_user_id),
+            page_id=page_id,
+            limit=limit,
+        )
+
+        if not success:
+            error_map = {
+                'not_a_member': (
+                    status.HTTP_403_FORBIDDEN,
+                    'You are not a member of this organization',
+                ),
+                'invalid_page_id': (
+                    status.HTTP_400_BAD_REQUEST,
+                    'Invalid page_id format',
+                ),
+            }
+            status_code, detail = error_map.get(
+                error_code, (status.HTTP_500_INTERNAL_SERVER_ERROR, 'An error occurred')
+            )
+            raise HTTPException(status_code=status_code, detail=detail)
+
+        if data is None:
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail='Failed to retrieve members',
+            )
+
+        return data
+
+    except HTTPException:
+        raise
+    except ValueError:
+        logger.exception('Invalid UUID format')
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail='Invalid organization ID format',
+        )
+    except Exception:
+        logger.exception('Error retrieving organization members')
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='Failed to retrieve members',
+        )
+
+
+@org_router.delete('/{org_id}/members/{user_id}')
+async def remove_org_member(
+    org_id: str,
+    user_id: str,
+    current_user_id: str = Depends(get_user_id),
+):
+    """Remove a member from an organization.
+
+    Only owners and admins can remove members:
+    - Owners can remove admins and regular users
+    - Admins can only remove regular users
+
+    Users cannot remove themselves. The last owner cannot be removed.
+    """
+    try:
+        success, error = await OrgMemberService.remove_org_member(
+            org_id=UUID(org_id),
+            target_user_id=UUID(user_id),
+            current_user_id=UUID(current_user_id),
+        )
+
+        if not success:
+            error_map = {
+                'not_a_member': (
+                    status.HTTP_403_FORBIDDEN,
+                    'You are not a member of this organization',
+                ),
+                'cannot_remove_self': (
+                    status.HTTP_403_FORBIDDEN,
+                    'Cannot remove yourself from an organization',
+                ),
+                'member_not_found': (
+                    status.HTTP_404_NOT_FOUND,
+                    'Member not found in this organization',
+                ),
+                'insufficient_permission': (
+                    status.HTTP_403_FORBIDDEN,
+                    'You do not have permission to remove this member',
+                ),
+                'cannot_remove_last_owner': (
+                    status.HTTP_400_BAD_REQUEST,
+                    'Cannot remove the last owner of an organization',
+                ),
+                'removal_failed': (
+                    status.HTTP_500_INTERNAL_SERVER_ERROR,
+                    'Failed to remove member',
+                ),
+            }
+            status_code, detail = error_map.get(
+                error, (status.HTTP_500_INTERNAL_SERVER_ERROR, 'An error occurred')
+            )
+            raise HTTPException(status_code=status_code, detail=detail)
+
+        return {'message': 'Member removed successfully'}
+
+    except HTTPException:
+        raise
+    except ValueError:
+        logger.exception('Invalid UUID format')
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail='Invalid organization or user ID format',
+        )
+    except Exception:
+        logger.exception('Error removing organization member')
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='Failed to remove member',
+        )
+
+
+@org_router.post(
+    '/{org_id}/switch', response_model=OrgResponse, status_code=status.HTTP_200_OK
+)
+async def switch_org(
+    org_id: UUID,
+    user_id: str = Depends(get_user_id),
+) -> OrgResponse:
+    """Switch to a different organization.
+
+    This endpoint allows authenticated users to switch their current active
+    organization. The user must be a member of the target organization.
+
+    Args:
+        org_id: Organization ID to switch to (UUID)
+        user_id: Authenticated user ID (injected by dependency)
+
+    Returns:
+        OrgResponse: The organization details that was switched to
+
+    Raises:
+        HTTPException: 422 if org_id is not a valid UUID (handled by FastAPI)
+        HTTPException: 403 if user is not a member of the organization
+        HTTPException: 404 if organization not found
+        HTTPException: 500 if switch fails
+    """
+    logger.info(
+        'Switching organization',
+        extra={
+            'user_id': user_id,
+            'org_id': str(org_id),
+        },
+    )
+
+    try:
+        # Use service layer to switch organization with membership validation
+        org = await OrgService.switch_org(
+            user_id=user_id,
+            org_id=org_id,
+        )
+
+        # Retrieve credits from LiteLLM for the new current org
+        credits = await OrgService.get_org_credits(user_id, org.id)
+
+        return OrgResponse.from_org(org, credits=credits, user_id=user_id)
+
+    except OrgNotFoundError as e:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=str(e),
+        )
+    except OrgAuthorizationError as e:
+        raise HTTPException(
+            status_code=status.HTTP_403_FORBIDDEN,
+            detail=str(e),
+        )
+    except OrgDatabaseError as e:
+        logger.error(
+            'Database operation failed during organization switch',
+            extra={'user_id': user_id, 'org_id': str(org_id), 'error': str(e)},
+        )
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='Failed to switch organization',
+        )
+    except Exception as e:
+        logger.exception(
+            'Unexpected error switching organization',
+            extra={'user_id': user_id, 'org_id': str(org_id), 'error': str(e)},
+        )
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='An unexpected error occurred',
+        )
+
+
+@org_router.patch('/{org_id}/members/{user_id}', response_model=OrgMemberResponse)
+async def update_org_member(
+    org_id: str,
+    user_id: str,
+    update_data: OrgMemberUpdate,
+    current_user_id: str = Depends(get_user_id),
+) -> OrgMemberResponse:
+    """Update a member's role in an organization.
+
+    Permission rules:
+    - Admins can change roles of regular members to Admin or Member
+    - Admins cannot modify other Admins or Owners
+    - Owners can change roles of Admins and Members to any role (Owner, Admin, Member)
+    - Owners cannot modify other Owners
+
+    Members cannot modify their own role. The last owner cannot be demoted.
+    """
+    try:
+        return await OrgMemberService.update_org_member(
+            org_id=UUID(org_id),
+            target_user_id=UUID(user_id),
+            current_user_id=UUID(current_user_id),
+            update_data=update_data,
+        )
+    except OrgMemberNotFoundError as e:
+        # Distinguish between requester not being a member vs target not found
+        if str(current_user_id) in str(e):
+            raise HTTPException(
+                status_code=status.HTTP_403_FORBIDDEN,
+                detail='You are not a member of this organization',
+            )
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail='Member not found in this organization',
+        )
+    except CannotModifySelfError:
+        raise HTTPException(
+            status_code=status.HTTP_403_FORBIDDEN,
+            detail='Cannot modify your own role',
+        )
+    except RoleNotFoundError:
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='Role configuration error',
+        )
+    except InvalidRoleError:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail='Invalid role specified',
+        )
+    except InsufficientPermissionError:
+        raise HTTPException(
+            status_code=status.HTTP_403_FORBIDDEN,
+            detail='You do not have permission to modify this member',
+        )
+    except LastOwnerError:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail='Cannot demote the last owner of an organization',
+        )
+    except MemberUpdateError:
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='Failed to update member',
+        )
+    except ValueError:
+        logger.exception('Invalid UUID format')
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail='Invalid organization or user ID format',
+        )
+    except Exception:
+        logger.exception('Error updating organization member')
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail='Failed to update member',
+        )

enterprise/server/routes/user.py

@@ -4,6 +4,7 @@ from fastapi import APIRouter, Depends, Query, status
 from fastapi.responses import JSONResponse
 from pydantic import SecretStr
 from server.auth.token_manager import TokenManager
+from utils.identity import resolve_display_name
 
 from openhands.integrations.provider import (
     PROVIDER_TOKEN_TYPE,
@@ -121,6 +122,8 @@ async def saas_get_user(
                 login=(user_info.get('preferred_username') if user_info else '') or '',
                 avatar_url='',
                 email=user_info.get('email') if user_info else None,
+                name=resolve_display_name(user_info) if user_info else None,
+                company=user_info.get('company') if user_info else None,
             ),
             user_info=user_info,
         )

enterprise/server/saas_nested_conversation_manager.py

@@ -516,11 +516,13 @@ class SaasNestedConversationManager(ConversationManager):
                     )
                     raise
 
-    def _get_mcp_config(self, user_id: str) -> MCPConfig | None:
+    async def _get_mcp_config(self, user_id: str) -> MCPConfig | None:
         api_key_store = ApiKeyStore.get_instance()
-        mcp_api_key = api_key_store.retrieve_mcp_api_key(user_id)
+        mcp_api_key = await api_key_store.retrieve_mcp_api_key(user_id)
         if not mcp_api_key:
-            mcp_api_key = api_key_store.create_api_key(user_id, 'MCP_API_KEY', None)
+            mcp_api_key = await api_key_store.create_api_key(
+                user_id, 'MCP_API_KEY', None
+            )
         if not mcp_api_key:
             return None
         web_host = os.environ.get('WEB_HOST', 'app.all-hands.dev')
@@ -547,7 +549,7 @@ class SaasNestedConversationManager(ConversationManager):
             'conversation_id': sid,
         }
 
-        mcp_config = self._get_mcp_config(user_id)
+        mcp_config = await self._get_mcp_config(user_id)
         if mcp_config:
             # Merge with any MCP config from settings
             if settings.mcp_config:
@@ -1137,6 +1139,71 @@ class SaasNestedConversationManager(ConversationManager):
         }
         update_conversation_metadata(conversation_id, metadata_content)
 
+    async def list_files(self, sid: str, path: str | None = None) -> list[str]:
+        """List files in the workspace for a conversation.
+
+        Delegates to the nested container's list-files endpoint.
+
+        Args:
+            sid: The session/conversation ID.
+            path: Optional path to list files from. If None, lists from workspace root.
+
+        Returns:
+            A list of file paths.
+
+        Raises:
+            ValueError: If the conversation is not running.
+            httpx.HTTPError: If there's an error communicating with the nested runtime.
+        """
+        runtime = await self._get_runtime(sid)
+        if runtime is None or runtime.get('status') != 'running':
+            raise ValueError(f'Conversation {sid} is not running')
+
+        nested_url = self._get_nested_url_for_runtime(runtime['runtime_id'], sid)
+        session_api_key = runtime.get('session_api_key')
+
+        return await self._fetch_list_files_from_nested(
+            sid, nested_url, session_api_key, path
+        )
+
+    async def select_file(self, sid: str, file: str) -> tuple[str | None, str | None]:
+        """Read a file from the workspace via nested container.
+
+        Raises:
+            ValueError: If the conversation is not running.
+            httpx.HTTPError: If there's an error communicating with the nested runtime.
+        """
+        runtime = await self._get_runtime(sid)
+        if runtime is None or runtime.get('status') != 'running':
+            raise ValueError(f'Conversation {sid} is not running')
+
+        nested_url = self._get_nested_url_for_runtime(runtime['runtime_id'], sid)
+        session_api_key = runtime.get('session_api_key')
+
+        return await self._fetch_select_file_from_nested(
+            sid, nested_url, session_api_key, file
+        )
+
+    async def upload_files(
+        self, sid: str, files: list[tuple[str, bytes]]
+    ) -> tuple[list[str], list[dict[str, str]]]:
+        """Upload files to the workspace via nested container.
+
+        Raises:
+            ValueError: If the conversation is not running.
+            httpx.HTTPError: If there's an error communicating with the nested runtime.
+        """
+        runtime = await self._get_runtime(sid)
+        if runtime is None or runtime.get('status') != 'running':
+            raise ValueError(f'Conversation {sid} is not running')
+
+        nested_url = self._get_nested_url_for_runtime(runtime['runtime_id'], sid)
+        session_api_key = runtime.get('session_api_key')
+
+        return await self._fetch_upload_files_to_nested(
+            sid, nested_url, session_api_key, files
+        )
+
 
 def _last_updated_at_key(conversation: ConversationMetadata) -> float:
     last_updated_at = conversation.last_updated_at

enterprise/server/services/org_member_service.py

@@ -0,0 +1,342 @@
+"""Service for managing organization members."""
+
+from uuid import UUID
+
+from server.constants import ROLE_ADMIN, ROLE_MEMBER, ROLE_OWNER
+from server.routes.org_models import (
+    CannotModifySelfError,
+    InsufficientPermissionError,
+    InvalidRoleError,
+    LastOwnerError,
+    MemberUpdateError,
+    MeResponse,
+    OrgMemberNotFoundError,
+    OrgMemberPage,
+    OrgMemberResponse,
+    OrgMemberUpdate,
+    RoleNotFoundError,
+)
+from storage.org_member_store import OrgMemberStore
+from storage.role_store import RoleStore
+from storage.user_store import UserStore
+
+from openhands.utils.async_utils import call_sync_from_async
+
+
+class OrgMemberService:
+    """Service for organization member operations."""
+
+    @staticmethod
+    def get_me(org_id: UUID, user_id: UUID) -> MeResponse:
+        """Get the current user's membership record for an organization.
+
+        Retrieves the authenticated user's role, status, email, and LLM override
+        fields (with masked API keys) within the specified organization.
+
+        Args:
+            org_id: Organization ID (UUID)
+            user_id: User ID (UUID)
+
+        Returns:
+            MeResponse: The user's membership data with masked API keys
+
+        Raises:
+            OrgMemberNotFoundError: If user is not a member of the organization
+            RoleNotFoundError: If the role associated with the member is not found
+        """
+        # Look up the user's membership in this org
+        org_member = OrgMemberStore.get_org_member(org_id, user_id)
+        if org_member is None:
+            raise OrgMemberNotFoundError(str(org_id), str(user_id))
+
+        # Resolve role name from role_id
+        role = RoleStore.get_role_by_id(org_member.role_id)
+        if role is None:
+            raise RoleNotFoundError(org_member.role_id)
+
+        # Get user email
+        user = UserStore.get_user_by_id(str(user_id))
+        email = user.email if user and user.email else ''
+
+        return MeResponse.from_org_member(org_member, role, email)
+
+    @staticmethod
+    async def get_org_members(
+        org_id: UUID,
+        current_user_id: UUID,
+        page_id: str | None = None,
+        limit: int = 100,
+    ) -> tuple[bool, str | None, OrgMemberPage | None]:
+        """Get organization members with authorization check.
+
+        Returns:
+            Tuple of (success, error_code, data). If success is True, error_code is None.
+        """
+        # Verify current user is a member of the organization
+        requester_membership = OrgMemberStore.get_org_member(org_id, current_user_id)
+        if not requester_membership:
+            return False, 'not_a_member', None
+
+        # Parse page_id to get offset (page_id is offset encoded as string)
+        offset = 0
+        if page_id is not None:
+            try:
+                offset = int(page_id)
+                if offset < 0:
+                    return False, 'invalid_page_id', None
+            except ValueError:
+                return False, 'invalid_page_id', None
+
+        # Call store to get paginated members
+        members, has_more = await OrgMemberStore.get_org_members_paginated(
+            org_id=org_id, offset=offset, limit=limit
+        )
+
+        # Transform data to response format
+        items = []
+        for member in members:
+            # Access user and role relationships (eagerly loaded)
+            user = member.user
+            role = member.role
+
+            items.append(
+                OrgMemberResponse(
+                    user_id=str(member.user_id),
+                    email=user.email if user else None,
+                    role_id=member.role_id,
+                    role_name=role.name if role else '',
+                    role_rank=role.rank if role else 0,
+                    status=member.status,
+                )
+            )
+
+        # Calculate next_page_id
+        next_page_id = None
+        if has_more:
+            next_page_id = str(offset + limit)
+
+        return True, None, OrgMemberPage(items=items, next_page_id=next_page_id)
+
+    @staticmethod
+    async def remove_org_member(
+        org_id: UUID,
+        target_user_id: UUID,
+        current_user_id: UUID,
+    ) -> tuple[bool, str | None]:
+        """Remove a member from an organization.
+
+        Returns:
+            Tuple of (success, error_message). If success is True, error_message is None.
+        """
+
+        def _remove_member():
+            # Get current user's membership in the org
+            requester_membership = OrgMemberStore.get_org_member(
+                org_id, current_user_id
+            )
+            if not requester_membership:
+                return False, 'not_a_member'
+
+            # Check if trying to remove self
+            if str(current_user_id) == str(target_user_id):
+                return False, 'cannot_remove_self'
+
+            # Get target user's membership
+            target_membership = OrgMemberStore.get_org_member(org_id, target_user_id)
+            if not target_membership:
+                return False, 'member_not_found'
+
+            requester_role = RoleStore.get_role_by_id(requester_membership.role_id)
+            target_role = RoleStore.get_role_by_id(target_membership.role_id)
+
+            if not requester_role or not target_role:
+                return False, 'role_not_found'
+
+            # Check permission based on roles
+            if not OrgMemberService._can_remove_member(
+                requester_role.name, target_role.name
+            ):
+                return False, 'insufficient_permission'
+
+            # Check if removing the last owner
+            if target_role.name == ROLE_OWNER:
+                if OrgMemberService._is_last_owner(org_id, target_user_id):
+                    return False, 'cannot_remove_last_owner'
+
+            # Perform the removal
+            success = OrgMemberStore.remove_user_from_org(org_id, target_user_id)
+            if not success:
+                return False, 'removal_failed'
+
+            return True, None
+
+        return await call_sync_from_async(_remove_member)
+
+    @staticmethod
+    async def update_org_member(
+        org_id: UUID,
+        target_user_id: UUID,
+        current_user_id: UUID,
+        update_data: OrgMemberUpdate,
+    ) -> OrgMemberResponse:
+        """Update a member's role in an organization.
+
+        Permission rules:
+        - Admins can change roles of users (rank > ADMIN_RANK) to Admin or User
+        - Admins cannot modify other Admins or Owners
+        - Owners can change roles of non-owners (rank > OWNER_RANK) to any role
+        - Owners cannot modify other Owners
+
+        Args:
+            org_id: Organization ID
+            target_user_id: User ID of the member to update
+            current_user_id: User ID of the requester
+            update_data: Update data containing fields to modify
+
+        Returns:
+            OrgMemberResponse: The updated member data
+
+        Raises:
+            OrgMemberNotFoundError: If requester or target is not a member
+            CannotModifySelfError: If trying to modify self
+            RoleNotFoundError: If role configuration is invalid
+            InvalidRoleError: If new_role_name is not a valid role
+            InsufficientPermissionError: If requester lacks permission
+            LastOwnerError: If trying to demote the last owner
+            MemberUpdateError: If update operation fails
+        """
+        new_role_name = update_data.role
+
+        def _update_member():
+            # Get current user's membership in the org
+            requester_membership = OrgMemberStore.get_org_member(
+                org_id, current_user_id
+            )
+            if not requester_membership:
+                raise OrgMemberNotFoundError(str(org_id), str(current_user_id))
+
+            # Check if trying to modify self
+            if str(current_user_id) == str(target_user_id):
+                raise CannotModifySelfError('modify')
+
+            # Get target user's membership
+            target_membership = OrgMemberStore.get_org_member(org_id, target_user_id)
+            if not target_membership:
+                raise OrgMemberNotFoundError(str(org_id), str(target_user_id))
+
+            # Get roles
+            requester_role = RoleStore.get_role_by_id(requester_membership.role_id)
+            target_role = RoleStore.get_role_by_id(target_membership.role_id)
+
+            if not requester_role:
+                raise RoleNotFoundError(requester_membership.role_id)
+            if not target_role:
+                raise RoleNotFoundError(target_membership.role_id)
+
+            # If no role change requested, return current state
+            if new_role_name is None:
+                user = UserStore.get_user_by_id(str(target_user_id))
+                return OrgMemberResponse(
+                    user_id=str(target_membership.user_id),
+                    email=user.email if user else None,
+                    role_id=target_membership.role_id,
+                    role_name=target_role.name,
+                    role_rank=target_role.rank,
+                    status=target_membership.status,
+                )
+
+            # Validate new role exists
+            new_role = RoleStore.get_role_by_name(new_role_name.lower())
+            if not new_role:
+                raise InvalidRoleError(new_role_name)
+
+            # Check permission to modify target
+            if not OrgMemberService._can_update_member_role(
+                requester_role.name, target_role.name, new_role.name
+            ):
+                raise InsufficientPermissionError(
+                    'You do not have permission to modify this member'
+                )
+
+            # Check if demoting the last owner
+            if (
+                target_role.name == ROLE_OWNER
+                and new_role.name != ROLE_OWNER
+                and OrgMemberService._is_last_owner(org_id, target_user_id)
+            ):
+                raise LastOwnerError('demote')
+
+            # Perform the update
+            updated_member = OrgMemberStore.update_user_role_in_org(
+                org_id, target_user_id, new_role.id
+            )
+            if not updated_member:
+                raise MemberUpdateError('Failed to update member')
+
+            # Get user email for response
+            user = UserStore.get_user_by_id(str(target_user_id))
+
+            return OrgMemberResponse(
+                user_id=str(updated_member.user_id),
+                email=user.email if user else None,
+                role_id=updated_member.role_id,
+                role_name=new_role.name,
+                role_rank=new_role.rank,
+                status=updated_member.status,
+            )
+
+        return await call_sync_from_async(_update_member)
+
+    @staticmethod
+    def _can_update_member_role(
+        requester_role_name: str, target_role_name: str, new_role_name: str
+    ) -> bool:
+        """Check if requester can change target's role to new_role.
+
+        Permission rules:
+        - Owners can modify admins and users, can set any role
+        - Owners cannot modify other owners
+        - Admins can only modify users
+        - Admins can only set admin or user roles (not owner)
+        """
+        is_requester_owner = requester_role_name == ROLE_OWNER
+        is_requester_admin = requester_role_name == ROLE_ADMIN
+        is_target_owner = target_role_name == ROLE_OWNER
+        is_target_admin = target_role_name == ROLE_ADMIN
+        is_new_role_owner = new_role_name == ROLE_OWNER
+
+        if is_requester_owner:
+            # Owners cannot modify other owners
+            if is_target_owner:
+                return False
+            # Owners can set any role (owner, admin, user)
+            return True
+        elif is_requester_admin:
+            # Admins cannot modify owners or other admins
+            if is_target_owner or is_target_admin:
+                return False
+            # Admins can only set admin or user roles (not owner)
+            return not is_new_role_owner
+        return False
+
+    @staticmethod
+    def _can_remove_member(requester_role_name: str, target_role_name: str) -> bool:
+        """Check if requester can remove target based on roles."""
+        if requester_role_name == ROLE_OWNER:
+            return True
+        elif requester_role_name == ROLE_ADMIN:
+            # Admins can only remove members (not owners or other admins)
+            return target_role_name == ROLE_MEMBER
+        return False
+
+    @staticmethod
+    def _is_last_owner(org_id: UUID, user_id: UUID) -> bool:
+        """Check if user is the last owner of the organization."""
+        members = OrgMemberStore.get_org_members(org_id)
+        owners = []
+        for m in members:
+            # Use role_id (column) instead of role (relationship) to avoid DetachedInstanceError
+            role = RoleStore.get_role_by_id(m.role_id)
+            if role and role.name == ROLE_OWNER:
+                owners.append(m)
+        return len(owners) == 1 and str(owners[0].user_id) == str(user_id)

enterprise/server/sharing/sql_shared_conversation_info_service.py

@@ -26,6 +26,7 @@ from server.sharing.shared_conversation_models import (
 )
 from sqlalchemy import select
 from sqlalchemy.ext.asyncio import AsyncSession
+from storage.stored_conversation_metadata_saas import StoredConversationMetadataSaas
 
 from openhands.app_server.app_conversation.sql_app_conversation_info_service import (
     StoredConversationMetadata,
@@ -57,7 +58,7 @@ class SQLSharedConversationInfoService(SharedConversationInfoService):
         include_sub_conversations: bool = False,
     ) -> SharedConversationPage:
         """Search for shared conversations."""
-        query = self._public_select()
+        query = self._public_select_with_saas_metadata()
 
         # Conditionally exclude sub-conversations based on the parameter
         if not include_sub_conversations:
@@ -104,14 +105,17 @@ class SQLSharedConversationInfoService(SharedConversationInfoService):
         query = query.limit(limit + 1)
 
         result = await self.db_session.execute(query)
-        rows = result.scalars().all()
+        rows = result.all()
 
         # Check if there are more results
         has_more = len(rows) > limit
         if has_more:
             rows = rows[:limit]
 
-        items = [self._to_shared_conversation(row) for row in rows]
+        items = [
+            self._to_shared_conversation(stored, saas_metadata=saas_metadata)
+            for stored, saas_metadata in rows
+        ]
 
         # Calculate next page ID
         next_page_id = None
@@ -152,17 +156,18 @@ class SQLSharedConversationInfoService(SharedConversationInfoService):
         self, conversation_id: UUID
     ) -> SharedConversation | None:
         """Get a single public conversation info, returning None if missing or not shared."""
-        query = self._public_select().where(
+        query = self._public_select_with_saas_metadata().where(
             StoredConversationMetadata.conversation_id == str(conversation_id)
         )
 
         result = await self.db_session.execute(query)
-        stored = result.scalar_one_or_none()
+        row = result.first()
 
-        if stored is None:
+        if row is None:
             return None
 
-        return self._to_shared_conversation(stored)
+        stored, saas_metadata = row
+        return self._to_shared_conversation(stored, saas_metadata=saas_metadata)
 
     def _public_select(self):
         """Create a select query that only returns public conversations."""
@@ -173,6 +178,25 @@ class SQLSharedConversationInfoService(SharedConversationInfoService):
         query = query.where(StoredConversationMetadata.public == True)  # noqa: E712
         return query
 
+    def _public_select_with_saas_metadata(self):
+        """Create a select query that returns public conversations with SAAS metadata.
+
+        This joins with conversation_metadata_saas to retrieve the user_id needed
+        for constructing the correct event storage path. Uses LEFT OUTER JOIN to
+        support conversations that may not have SAAS metadata (e.g., in tests).
+        """
+        query = (
+            select(StoredConversationMetadata, StoredConversationMetadataSaas)
+            .outerjoin(
+                StoredConversationMetadataSaas,
+                StoredConversationMetadata.conversation_id
+                == StoredConversationMetadataSaas.conversation_id,
+            )
+            .where(StoredConversationMetadata.conversation_version == 'V1')
+            .where(StoredConversationMetadata.public == True)  # noqa: E712
+        )
+        return query
+
     def _apply_filters(
         self,
         query,
@@ -211,9 +235,16 @@ class SQLSharedConversationInfoService(SharedConversationInfoService):
     def _to_shared_conversation(
         self,
         stored: StoredConversationMetadata,
+        saas_metadata: StoredConversationMetadataSaas | None = None,
         sub_conversation_ids: list[UUID] | None = None,
     ) -> SharedConversation:
-        """Convert StoredConversationMetadata to SharedConversation."""
+        """Convert StoredConversationMetadata to SharedConversation.
+
+        Args:
+            stored: The base conversation metadata from conversation_metadata table.
+            saas_metadata: Optional SAAS metadata containing user_id and org_id.
+            sub_conversation_ids: Optional list of sub-conversation IDs.
+        """
         # V1 conversations should always have a sandbox_id
         sandbox_id = stored.sandbox_id
         assert sandbox_id is not None
@@ -239,9 +270,16 @@ class SQLSharedConversationInfoService(SharedConversationInfoService):
         created_at = self._fix_timezone(stored.created_at)
         updated_at = self._fix_timezone(stored.last_updated_at)
 
+        # Get user_id from SAAS metadata if available
+        created_by_user_id = (
+            str(saas_metadata.user_id)
+            if saas_metadata and saas_metadata.user_id
+            else None
+        )
+
         return SharedConversation(
             id=UUID(stored.conversation_id),
-            created_by_user_id=None,  # user_id is no longer stored in conversation metadata
+            created_by_user_id=created_by_user_id,
             sandbox_id=stored.sandbox_id,
             selected_repository=stored.selected_repository,
             selected_branch=stored.selected_branch,

enterprise/server/utils/saas_app_conversation_info_injector.py

@@ -233,7 +233,13 @@ class SaasSQLAppConversationInfoService(SQLAppConversationInfoService):
         result = result_set.first()
         if result:
             stored_metadata, saas_metadata = result
-            return self._to_info_with_user_id(stored_metadata, saas_metadata)
+            # Fetch sub-conversation IDs
+            sub_conversation_ids = await self.get_sub_conversation_ids(conversation_id)
+            return self._to_info_with_user_id(
+                stored_metadata,
+                saas_metadata,
+                sub_conversation_ids=sub_conversation_ids,
+            )
         return None
 
     async def batch_get_app_conversation_info(
@@ -262,8 +268,16 @@ class SaasSQLAppConversationInfoService(SQLAppConversationInfoService):
         for conversation_id in conversation_id_strs:
             if conversation_id in info_by_id:
                 stored_metadata, saas_metadata = info_by_id[conversation_id]
+                # Fetch sub-conversation IDs for each conversation
+                sub_conversation_ids = await self.get_sub_conversation_ids(
+                    UUID(conversation_id)
+                )
                 results.append(
-                    self._to_info_with_user_id(stored_metadata, saas_metadata)
+                    self._to_info_with_user_id(
+                        stored_metadata,
+                        saas_metadata,
+                        sub_conversation_ids=sub_conversation_ids,
+                    )
                 )
             else:
                 results.append(None)
@@ -316,10 +330,11 @@ class SaasSQLAppConversationInfoService(SQLAppConversationInfoService):
         self,
         stored: StoredConversationMetadata,
         saas_metadata: StoredConversationMetadataSaas,
+        sub_conversation_ids: list[UUID] | None = None,
     ) -> AppConversationInfo:
         """Convert stored metadata to AppConversationInfo with user_id from SAAS metadata."""
         # Use the base _to_info method to get the basic info
-        info = self._to_info(stored)
+        info = self._to_info(stored, sub_conversation_ids=sub_conversation_ids)
 
         # Override the created_by_user_id with the user_id from SAAS metadata
         info.created_by_user_id = (

enterprise/storage/api_key_store.py

@@ -12,6 +12,7 @@ from storage.database import session_maker
 from storage.user_store import UserStore
 
 from openhands.core.logger import openhands_logger as logger
+from openhands.utils.async_utils import call_sync_from_async
 
 
 @dataclass
@@ -26,7 +27,7 @@ class ApiKeyStore:
         random_part = ''.join(secrets.choice(alphabet) for _ in range(length))
         return f'{self.API_KEY_PREFIX}{random_part}'
 
-    def create_api_key(
+    async def create_api_key(
         self, user_id: str, name: str | None = None, expires_at: datetime | None = None
     ) -> str:
         """Create a new API key for a user.
@@ -40,8 +41,23 @@ class ApiKeyStore:
             The generated API key
         """
         api_key = self.generate_api_key()
-        user = UserStore.get_user_by_id(user_id)
+        user = await UserStore.get_user_by_id_async(user_id)
         org_id = user.current_org_id
+        await call_sync_from_async(
+            self._store_api_key, user_id, org_id, api_key, name, expires_at
+        )
+
+        return api_key
+
+    def _store_api_key(
+        self,
+        user_id: str,
+        org_id: str,
+        api_key: str,
+        name: str | None,
+        expires_at: datetime | None = None,
+    ) -> None:
+        """Store an existing API key in the database."""
         with self.session_maker() as session:
             key_record = ApiKey(
                 key=api_key,
@@ -53,8 +69,6 @@ class ApiKeyStore:
             session.add(key_record)
             session.commit()
 
-        return api_key
-
     def validate_api_key(self, api_key: str) -> str | None:
         """Validate an API key and return the associated user_id if valid."""
         now = datetime.now(UTC)
@@ -112,10 +126,13 @@ class ApiKeyStore:
 
             return True
 
-    def list_api_keys(self, user_id: str) -> list[dict]:
+    async def list_api_keys(self, user_id: str) -> list[dict]:
         """List all API keys for a user."""
-        user = UserStore.get_user_by_id(user_id)
+        user = await UserStore.get_user_by_id_async(user_id)
         org_id = user.current_org_id
+        return await call_sync_from_async(self._list_api_keys_from_db, user_id, org_id)
+
+    def _list_api_keys_from_db(self, user_id: str, org_id: str) -> list[ApiKey]:
         with self.session_maker() as session:
             keys = (
                 session.query(ApiKey)
@@ -136,9 +153,14 @@ class ApiKeyStore:
                 if 'MCP_API_KEY' != key.name
             ]
 
-    def retrieve_mcp_api_key(self, user_id: str) -> str | None:
-        user = UserStore.get_user_by_id(user_id)
+    async def retrieve_mcp_api_key(self, user_id: str) -> str | None:
+        user = await UserStore.get_user_by_id_async(user_id)
         org_id = user.current_org_id
+        return await call_sync_from_async(
+            self._retrieve_mcp_api_key_from_db, user_id, org_id
+        )
+
+    def _retrieve_mcp_api_key_from_db(self, user_id: str, org_id: str) -> str | None:
         with self.session_maker() as session:
             keys: list[ApiKey] = (
                 session.query(ApiKey)

enterprise/storage/encrypt_utils.py

@@ -98,6 +98,29 @@ def decrypt_legacy_value(value: str | SecretStr) -> str:
         return get_fernet().decrypt(b64decode(value.encode())).decode()
 
 
+def encrypt_legacy_model(encrypt_keys: list, model_instance) -> dict:
+    return encrypt_legacy_kwargs(encrypt_keys, model_to_kwargs(model_instance))
+
+
+def encrypt_legacy_kwargs(encrypt_keys: list, kwargs: dict) -> dict:
+    for key, value in kwargs.items():
+        if value is None:
+            continue
+        if key in encrypt_keys:
+            value = encrypt_legacy_value(value)
+            kwargs[key] = value
+    return kwargs
+
+
+def encrypt_legacy_value(value: str | SecretStr) -> str:
+    if isinstance(value, SecretStr):
+        return b64encode(
+            get_fernet().encrypt(value.get_secret_value().encode())
+        ).decode()
+    else:
+        return b64encode(get_fernet().encrypt(value.encode())).decode()
+
+
 def get_fernet():
     global _fernet
     if _fernet is None:

enterprise/storage/lite_llm_manager.py

@@ -18,18 +18,29 @@ from server.constants import (
     get_default_litellm_model,
 )
 from server.logger import logger
+from storage.encrypt_utils import decrypt_legacy_value
 from storage.user_settings import UserSettings
 
 from openhands.server.settings import Settings
 from openhands.utils.http_session import httpx_verify_option
 
-# Timeout in seconds for BYOR key verification requests to LiteLLM
-BYOR_KEY_VERIFICATION_TIMEOUT = 5.0
+# Timeout in seconds for key verification requests to LiteLLM
+KEY_VERIFICATION_TIMEOUT = 5.0
 
 # A very large number to represent "unlimited" until LiteLLM fixes their unlimited update bug.
 UNLIMITED_BUDGET_SETTING = 1000000000.0
 
 
+def get_openhands_cloud_key_alias(keycloak_user_id: str, org_id: str) -> str:
+    """Generate the key alias for OpenHands Cloud managed keys."""
+    return f'OpenHands Cloud - user {keycloak_user_id} - org {org_id}'
+
+
+def get_byor_key_alias(keycloak_user_id: str, org_id: str) -> str:
+    """Generate the key alias for BYOR (Bring Your Own Runtime) keys."""
+    return f'BYOR Key - user {keycloak_user_id}, org {org_id}'
+
+
 class LiteLlmManager:
     """Manage LiteLLM interactions."""
 
@@ -78,7 +89,7 @@ class LiteLlmManager:
                     client,
                     keycloak_user_id,
                     org_id,
-                    f'OpenHands Cloud - user {keycloak_user_id} - org {org_id}',
+                    get_openhands_cloud_key_alias(keycloak_user_id, org_id),
                     None,
                 )
 
@@ -96,7 +107,7 @@ class LiteLlmManager:
         user_settings: UserSettings,
     ) -> UserSettings | None:
         logger.info(
-            'SettingsStore:umigrate_lite_llm_entries:start',
+            'LiteLlmManager:migrate_lite_llm_entries:start',
             extra={'org_id': org_id, 'user_id': keycloak_user_id},
         )
         if LITE_LLM_API_KEY is None or LITE_LLM_API_URL is None:
@@ -114,8 +125,24 @@ class LiteLlmManager:
                 if not user_json:
                     return None
                 user_info = user_json['user_info']
-                max_budget = user_info.get('max_budget', 0.0)
-                spend = user_info.get('spend', 0.0)
+
+                # Log original user values before any modifications for debugging
+                original_max_budget = user_info.get('max_budget')
+                original_spend = user_info.get('spend')
+                logger.info(
+                    'LiteLlmManager:migrate_lite_llm_entries:original_user_values',
+                    extra={
+                        'org_id': org_id,
+                        'user_id': keycloak_user_id,
+                        'original_max_budget': original_max_budget,
+                        'original_spend': original_spend,
+                    },
+                )
+
+                max_budget = (
+                    original_max_budget if original_max_budget is not None else 0.0
+                )
+                spend = original_spend if original_spend is not None else 0.0
                 # In upgrade to V4, we no longer use billing margin, but instead apply this directly
                 # in litellm. The default billing marign was 2 before this (hence the magic numbers below)
                 if (
@@ -136,39 +163,263 @@ class LiteLlmManager:
                     max_budget *= billing_margin
                     spend *= billing_margin
 
-                if not max_budget:
-                    # if max_budget is None, then we've already migrated the User
+                # Check if max_budget is None (not 0.0) or set to unlimited to determine if already migrated
+                # A user with max_budget=0.0 is different from max_budget=None
+                if (
+                    original_max_budget is None
+                    or original_max_budget == UNLIMITED_BUDGET_SETTING
+                ):
+                    # if max_budget is None or UNLIMITED, then we've already migrated the User
+                    logger.info(
+                        'LiteLlmManager:migrate_lite_llm_entries:already_migrated',
+                        extra={
+                            'org_id': org_id,
+                            'user_id': keycloak_user_id,
+                            'original_max_budget': original_max_budget,
+                        },
+                    )
                     return None
                 credits = max(max_budget - spend, 0.0)
 
+                # Log calculated migration values before performing updates
+                logger.info(
+                    'LiteLlmManager:migrate_lite_llm_entries:calculated_values',
+                    extra={
+                        'org_id': org_id,
+                        'user_id': keycloak_user_id,
+                        'adjusted_max_budget': max_budget,
+                        'adjusted_spend': spend,
+                        'calculated_credits': credits,
+                        'new_user_max_budget': UNLIMITED_BUDGET_SETTING,
+                    },
+                )
+
+                logger.debug(
+                    'LiteLlmManager:migrate_lite_llm_entries:create_team',
+                    extra={'org_id': org_id, 'user_id': keycloak_user_id},
+                )
                 await LiteLlmManager._create_team(
                     client, keycloak_user_id, org_id, credits
                 )
 
+                logger.debug(
+                    'LiteLlmManager:migrate_lite_llm_entries:update_user',
+                    extra={'org_id': org_id, 'user_id': keycloak_user_id},
+                )
                 await LiteLlmManager._update_user(
                     client, keycloak_user_id, max_budget=UNLIMITED_BUDGET_SETTING
                 )
 
+                logger.debug(
+                    'LiteLlmManager:migrate_lite_llm_entries:add_user_to_team',
+                    extra={'org_id': org_id, 'user_id': keycloak_user_id},
+                )
                 await LiteLlmManager._add_user_to_team(
                     client, keycloak_user_id, org_id, credits
                 )
 
-                if user_settings.llm_api_key:
-                    await LiteLlmManager._update_key(
-                        client,
-                        keycloak_user_id,
-                        user_settings.llm_api_key,
-                        team_id=org_id,
+                logger.debug(
+                    'LiteLlmManager:migrate_lite_llm_entries:update_user_keys',
+                    extra={'org_id': org_id, 'user_id': keycloak_user_id},
+                )
+                await LiteLlmManager._update_user_keys(
+                    client,
+                    keycloak_user_id,
+                    team_id=org_id,
+                )
+
+                # Check if the database key exists in LiteLLM
+                # If not, generate a new key to prevent verification failures later
+                db_key = None
+                if (
+                    user_settings
+                    and user_settings.llm_api_key
+                    and user_settings.llm_base_url == LITE_LLM_API_URL
+                ):
+                    db_key = user_settings.llm_api_key
+                    if hasattr(db_key, 'get_secret_value'):
+                        db_key = db_key.get_secret_value()
+
+                if db_key:
+                    # Verify the database key exists in LiteLLM
+                    key_valid = await LiteLlmManager.verify_key(
+                        db_key, keycloak_user_id
+                    )
+                    if not key_valid:
+                        logger.warning(
+                            'LiteLlmManager:migrate_lite_llm_entries:db_key_not_in_litellm',
+                            extra={
+                                'org_id': org_id,
+                                'user_id': keycloak_user_id,
+                                'key_prefix': db_key[:10] + '...'
+                                if len(db_key) > 10
+                                else db_key,
+                            },
+                        )
+                        # Generate a new key for the user
+                        new_key = await LiteLlmManager._generate_key(
+                            client,
+                            keycloak_user_id,
+                            org_id,
+                            get_openhands_cloud_key_alias(keycloak_user_id, org_id),
+                            None,
+                        )
+                        if new_key:
+                            logger.info(
+                                'LiteLlmManager:migrate_lite_llm_entries:generated_new_key',
+                                extra={'org_id': org_id, 'user_id': keycloak_user_id},
+                            )
+                            # Update user_settings with the new key so it gets stored in org_member
+                            user_settings.llm_api_key = SecretStr(new_key)
+                            user_settings.llm_api_key_for_byor = SecretStr(new_key)
+
+        logger.info(
+            'LiteLlmManager:migrate_lite_llm_entries:complete',
+            extra={'org_id': org_id, 'user_id': keycloak_user_id},
+        )
+        return user_settings
+
+    @staticmethod
+    async def downgrade_entries(
+        org_id: str,
+        keycloak_user_id: str,
+        user_settings: UserSettings,
+    ) -> UserSettings | None:
+        """Downgrade a migrated user's LiteLLM entries back to the pre-migration state.
+
+        This reverses the migrate_entries operation:
+        1. Get the user max budget from their org team in litellm
+        2. Set the max budget in the user in litellm (restore from team)
+        3. Add the user back to the default team in litellm
+        4. Update keys to remove org team association
+        5. Remove the user from their org team in litellm
+        6. Delete the user org team in litellm
+
+        Note: The database changes (already_migrated flag, org/org_member deletion)
+        should be handled separately by the caller.
+
+        Args:
+            org_id: The organization ID (which is also the team_id in litellm)
+            keycloak_user_id: The user's Keycloak ID
+            user_settings: The user's settings object
+
+        Returns:
+            The user_settings if downgrade was successful, None otherwise
+        """
+        logger.info(
+            'LiteLlmManager:downgrade_entries:start',
+            extra={'org_id': org_id, 'user_id': keycloak_user_id},
+        )
+        if LITE_LLM_API_KEY is None or LITE_LLM_API_URL is None:
+            logger.warning('LiteLLM API configuration not found')
+            return None
+
+        local_deploy = os.environ.get('LOCAL_DEPLOYMENT', None)
+        if not local_deploy:
+            async with httpx.AsyncClient(
+                headers={
+                    'x-goog-api-key': LITE_LLM_API_KEY,
+                }
+            ) as client:
+                # Step 1: Get the team info to retrieve the budget
+                logger.debug(
+                    'LiteLlmManager:downgrade_entries:get_team',
+                    extra={'org_id': org_id, 'user_id': keycloak_user_id},
+                )
+                team_info = await LiteLlmManager._get_team(client, org_id)
+                if not team_info:
+                    logger.error(
+                        'LiteLlmManager:downgrade_entries:team_not_found',
+                        extra={'org_id': org_id, 'user_id': keycloak_user_id},
+                    )
+                    return None
+
+                # Get team budget (max_budget) and spend to calculate current credits
+                team_data = team_info.get('team_info', {})
+                max_budget = team_data.get('max_budget', 0.0)
+                spend = team_data.get('spend', 0.0)
+
+                # Get user membership info for budget in team
+                user_membership = await LiteLlmManager._get_user_team_info(
+                    client, keycloak_user_id, org_id
+                )
+                if user_membership:
+                    # Use user's budget in team if available
+                    user_max_budget_in_team = user_membership.get('max_budget_in_team')
+                    user_spend_in_team = user_membership.get('spend', 0.0)
+                    if user_max_budget_in_team is not None:
+                        max_budget = user_max_budget_in_team
+                        spend = user_spend_in_team
+
+                # Calculate total budget to restore (credits + spend = max_budget)
+                # We restore the full max_budget that was on the team/user-in-team
+                restored_budget = max_budget if max_budget else 0.0
+
+                logger.debug(
+                    'LiteLlmManager:downgrade_entries:budget_info',
+                    extra={
+                        'org_id': org_id,
+                        'user_id': keycloak_user_id,
+                        'max_budget': max_budget,
+                        'spend': spend,
+                        'restored_budget': restored_budget,
+                    },
+                )
+
+                # Step 2: Update user to set their max_budget back from unlimited
+                logger.debug(
+                    'LiteLlmManager:downgrade_entries:update_user',
+                    extra={'org_id': org_id, 'user_id': keycloak_user_id},
+                )
+                await LiteLlmManager._update_user(
+                    client, keycloak_user_id, max_budget=restored_budget, spend=spend
+                )
+
+                # Step 3: Add user back to the default team
+                if LITE_LLM_TEAM_ID:
+                    logger.debug(
+                        'LiteLlmManager:downgrade_entries:add_to_default_team',
+                        extra={
+                            'org_id': org_id,
+                            'user_id': keycloak_user_id,
+                            'default_team_id': LITE_LLM_TEAM_ID,
+                        },
+                    )
+                    await LiteLlmManager._add_user_to_team(
+                        client, keycloak_user_id, LITE_LLM_TEAM_ID, restored_budget
                     )
 
-                if user_settings.llm_api_key_for_byor:
-                    await LiteLlmManager._update_key(
-                        client,
-                        keycloak_user_id,
-                        user_settings.llm_api_key_for_byor,
-                        team_id=org_id,
-                    )
+                # Step 4: Update all user keys to remove org team association (set team_id to default)
+                logger.debug(
+                    'LiteLlmManager:downgrade_entries:update_user_keys',
+                    extra={'org_id': org_id, 'user_id': keycloak_user_id},
+                )
+                await LiteLlmManager._update_user_keys(
+                    client,
+                    keycloak_user_id,
+                    team_id=LITE_LLM_TEAM_ID,
+                )
 
+                # Step 5: Remove user from their org team
+                logger.debug(
+                    'LiteLlmManager:downgrade_entries:remove_from_org_team',
+                    extra={'org_id': org_id, 'user_id': keycloak_user_id},
+                )
+                await LiteLlmManager._remove_user_from_team(
+                    client, keycloak_user_id, org_id
+                )
+
+                # Step 6: Delete the org team
+                logger.debug(
+                    'LiteLlmManager:downgrade_entries:delete_team',
+                    extra={'org_id': org_id, 'user_id': keycloak_user_id},
+                )
+                await LiteLlmManager._delete_team(client, org_id)
+
+        logger.info(
+            'LiteLlmManager:downgrade_entries:complete',
+            extra={'org_id': org_id, 'user_id': keycloak_user_id},
+        )
         return user_settings
 
     @staticmethod
@@ -424,6 +675,13 @@ class LiteLlmManager:
             logger.warning('LiteLLM API configuration not found')
             return
 
+        try:
+            # Sometimes the key we get is encrypted - attempt to decrypt.
+            key = decrypt_legacy_value(key)
+        except Exception:
+            # The key was not encrypted
+            pass
+
         payload = {
             'key': key,
         }
@@ -440,6 +698,7 @@ class LiteLlmManager:
                     'invalid_litellm_key_during_update',
                     extra={
                         'user_id': keycloak_user_id,
+                        'text': response.text,
                     },
                 )
                 return
@@ -453,6 +712,77 @@ class LiteLlmManager:
             )
         response.raise_for_status()
 
+    @staticmethod
+    async def _get_user_keys(
+        client: httpx.AsyncClient,
+        keycloak_user_id: str,
+    ) -> list[str]:
+        """Get all keys for a user from LiteLLM.
+
+        Args:
+            client: The HTTP client to use for the request
+            keycloak_user_id: The user's Keycloak ID
+
+        Returns:
+            A list of key strings belonging to the user
+        """
+        if LITE_LLM_API_KEY is None or LITE_LLM_API_URL is None:
+            logger.warning('LiteLLM API configuration not found')
+            return []
+
+        response = await client.get(
+            f'{LITE_LLM_API_URL}/key/list',
+            params={'user_id': keycloak_user_id},
+        )
+
+        if not response.is_success:
+            logger.error(
+                'error_getting_user_keys',
+                extra={
+                    'status_code': response.status_code,
+                    'text': response.text,
+                    'user_id': keycloak_user_id,
+                },
+            )
+            return []
+
+        response_json = response.json()
+        keys = response_json.get('keys', [])
+        logger.debug(
+            'LiteLlmManager:_get_user_keys:keys_retrieved',
+            extra={
+                'user_id': keycloak_user_id,
+                'key_count': len(keys),
+            },
+        )
+        return keys
+
+    @staticmethod
+    async def _update_user_keys(
+        client: httpx.AsyncClient,
+        keycloak_user_id: str,
+        **kwargs,
+    ):
+        """Update all keys belonging to a user with the given parameters.
+
+        Args:
+            client: The HTTP client to use for the request
+            keycloak_user_id: The user's Keycloak ID
+            **kwargs: Parameters to update on each key (e.g., team_id)
+        """
+        keys = await LiteLlmManager._get_user_keys(client, keycloak_user_id)
+
+        logger.debug(
+            'LiteLlmManager:_update_user_keys:updating_keys',
+            extra={
+                'user_id': keycloak_user_id,
+                'key_count': len(keys),
+            },
+        )
+
+        for key in keys:
+            await LiteLlmManager._update_key(client, keycloak_user_id, key, **kwargs)
+
     @staticmethod
     async def _delete_user(
         client: httpx.AsyncClient,
@@ -613,6 +943,45 @@ class LiteLlmManager:
             )
         response.raise_for_status()
 
+    @staticmethod
+    async def _remove_user_from_team(
+        client: httpx.AsyncClient,
+        keycloak_user_id: str,
+        team_id: str,
+    ):
+        if LITE_LLM_API_KEY is None or LITE_LLM_API_URL is None:
+            logger.warning('LiteLLM API configuration not found')
+            return
+        response = await client.post(
+            f'{LITE_LLM_API_URL}/team/member_delete',
+            json={
+                'team_id': team_id,
+                'user_id': keycloak_user_id,
+            },
+        )
+        if not response.is_success:
+            if response.status_code == 404:
+                # User not in team, that's fine for downgrade
+                logger.info(
+                    'User not in team during removal',
+                    extra={'user_id': keycloak_user_id, 'team_id': team_id},
+                )
+                return
+            logger.error(
+                'error_removing_litellm_user_from_team',
+                extra={
+                    'status_code': response.status_code,
+                    'text': response.text,
+                    'user_id': keycloak_user_id,
+                    'team_id': team_id,
+                },
+            )
+        response.raise_for_status()
+        logger.info(
+            'LiteLlmManager:_remove_user_from_team:user_removed',
+            extra={'user_id': keycloak_user_id, 'team_id': team_id},
+        )
+
     @staticmethod
     async def _generate_key(
         client: httpx.AsyncClient,
@@ -685,7 +1054,7 @@ class LiteLlmManager:
         try:
             async with httpx.AsyncClient(
                 verify=httpx_verify_option(),
-                timeout=BYOR_KEY_VERIFICATION_TIMEOUT,
+                timeout=KEY_VERIFICATION_TIMEOUT,
             ) as client:
                 # Make a lightweight request to verify the key
                 # Using /v1/models endpoint as it's lightweight and requires authentication
@@ -699,7 +1068,7 @@ class LiteLlmManager:
                 # Only 200 status code indicates valid key
                 if response.status_code == 200:
                     logger.debug(
-                        'BYOR key verification successful',
+                        'Key verification successful',
                         extra={'user_id': user_id},
                     )
                     return True
@@ -707,7 +1076,7 @@ class LiteLlmManager:
                 # All other status codes (401, 403, 500, etc.) are treated as invalid
                 # This includes authentication errors and server errors
                 logger.warning(
-                    'BYOR key verification failed - treating as invalid',
+                    'Key verification failed - treating as invalid',
                     extra={
                         'user_id': user_id,
                         'status_code': response.status_code,
@@ -720,7 +1089,7 @@ class LiteLlmManager:
             # Any exception (timeout, network error, etc.) means we can't verify
             # Return False to trigger regeneration rather than returning potentially invalid key
             logger.warning(
-                'BYOR key verification error - treating as invalid to ensure key validity',
+                'Key verification error - treating as invalid to ensure key validity',
                 extra={
                     'user_id': user_id,
                     'error': str(e),
@@ -764,6 +1133,103 @@ class LiteLlmManager:
             'key_spend': key_info.get('spend'),
         }
 
+    @staticmethod
+    async def _get_all_keys_for_user(
+        client: httpx.AsyncClient,
+        keycloak_user_id: str,
+    ) -> list[dict]:
+        """Get all keys for a user from LiteLLM.
+
+        Returns a list of key info dictionaries containing:
+        - token: the key value (hashed or partial)
+        - key_alias: the alias for the key
+        - key_name: the name of the key
+        - spend: the amount spent on this key
+        - max_budget: the max budget for this key
+        - team_id: the team the key belongs to
+        - metadata: any metadata associated with the key
+
+        Returns an empty list if no keys found or on error.
+        """
+        if LITE_LLM_API_KEY is None or LITE_LLM_API_URL is None:
+            logger.warning('LiteLLM API configuration not found')
+            return []
+
+        try:
+            response = await client.get(
+                f'{LITE_LLM_API_URL}/user/info?user_id={keycloak_user_id}',
+                headers={'x-goog-api-key': LITE_LLM_API_KEY},
+            )
+            response.raise_for_status()
+            user_json = response.json()
+            # The user/info endpoint returns keys in the 'keys' field
+            return user_json.get('keys', [])
+        except Exception as e:
+            logger.warning(
+                'LiteLlmManager:_get_all_keys_for_user:error',
+                extra={
+                    'user_id': keycloak_user_id,
+                    'error': str(e),
+                },
+            )
+            return []
+
+    @staticmethod
+    async def _verify_existing_key(
+        client: httpx.AsyncClient,
+        key_value: str,
+        keycloak_user_id: str,
+        org_id: str,
+        openhands_type: bool = False,
+    ) -> bool:
+        """Check if an existing key exists for the user/org in LiteLLM.
+
+        Verifies the provided key_value matches a key registered in LiteLLM for
+        the given user and organization. For openhands_type=True, looks for keys
+        with metadata type='openhands' and matching team_id. For openhands_type=False,
+        looks for keys with matching alias and team_id.
+
+        Returns True if the key is found and valid, False otherwise.
+        """
+        found = False
+        keys = await LiteLlmManager._get_all_keys_for_user(client, keycloak_user_id)
+        for key_info in keys:
+            metadata = key_info.get('metadata') or {}
+            team_id = key_info.get('team_id')
+            key_alias = key_info.get('key_alias')
+            token = None
+            if (
+                openhands_type
+                and metadata.get('type') == 'openhands'
+                and team_id == org_id
+            ):
+                # Found an existing OpenHands key for this org
+                key_name = key_info.get('key_name')
+                token = key_name[-4:] if key_name else None  # last 4 digits of key
+                if token and key_value.endswith(
+                    token
+                ):  # check if this is our current key
+                    found = True
+                    break
+            if (
+                not openhands_type
+                and team_id == org_id
+                and (
+                    key_alias == get_openhands_cloud_key_alias(keycloak_user_id, org_id)
+                    or key_alias == get_byor_key_alias(keycloak_user_id, org_id)
+                )
+            ):
+                # Found an existing key for this org (regardless of type)
+                key_name = key_info.get('key_name')
+                token = key_name[-4:] if key_name else None  # last 4 digits of key
+                if token and key_value.endswith(
+                    token
+                ):  # check if this is our current key
+                    found = True
+                    break
+
+        return found
+
     @staticmethod
     async def _delete_key_by_alias(
         client: httpx.AsyncClient,
@@ -856,8 +1322,13 @@ class LiteLlmManager:
     delete_user = staticmethod(with_http_client(_delete_user))
     delete_team = staticmethod(with_http_client(_delete_team))
     add_user_to_team = staticmethod(with_http_client(_add_user_to_team))
+    remove_user_from_team = staticmethod(with_http_client(_remove_user_from_team))
     get_user_team_info = staticmethod(with_http_client(_get_user_team_info))
     update_user_in_team = staticmethod(with_http_client(_update_user_in_team))
     generate_key = staticmethod(with_http_client(_generate_key))
     get_key_info = staticmethod(with_http_client(_get_key_info))
+    verify_existing_key = staticmethod(with_http_client(_verify_existing_key))
     delete_key = staticmethod(with_http_client(_delete_key))
+    get_user_keys = staticmethod(with_http_client(_get_user_keys))
+    delete_key_by_alias = staticmethod(with_http_client(_delete_key_by_alias))
+    update_user_keys = staticmethod(with_http_client(_update_user_keys))

enterprise/storage/org.py

@@ -46,6 +46,7 @@ class Org(Base):  # type: ignore
     v1_enabled = Column(Boolean, nullable=True)
     conversation_expiration = Column(Integer, nullable=True)
     condenser_max_size = Column(Integer, nullable=True)
+    byor_export_enabled = Column(Boolean, nullable=False, default=False)
 
     # Relationships
     org_members = relationship('OrgMember', back_populates='org')

enterprise/storage/org_member_store.py

@@ -5,7 +5,9 @@ Store class for managing organization-member relationships.
 from typing import Optional
 from uuid import UUID
 
-from storage.database import session_maker
+from sqlalchemy import select
+from sqlalchemy.orm import joinedload
+from storage.database import a_session_maker, session_maker
 from storage.org_member import OrgMember
 from storage.user_settings import UserSettings
 
@@ -38,7 +40,7 @@ class OrgMemberStore:
             return org_member
 
     @staticmethod
-    def get_org_member(org_id: UUID, user_id: int) -> Optional[OrgMember]:
+    def get_org_member(org_id: UUID, user_id: UUID) -> Optional[OrgMember]:
         """Get organization-user relationship."""
         with session_maker() as session:
             return (
@@ -48,7 +50,18 @@ class OrgMemberStore:
             )
 
     @staticmethod
-    def get_user_orgs(user_id: int) -> list[OrgMember]:
+    async def get_org_member_async(org_id: UUID, user_id: UUID) -> Optional[OrgMember]:
+        """Get organization-user relationship."""
+        async with a_session_maker() as session:
+            result = await session.execute(
+                select(OrgMember).filter(
+                    OrgMember.org_id == org_id, OrgMember.user_id == user_id
+                )
+            )
+            return result.scalars().first()
+
+    @staticmethod
+    def get_user_orgs(user_id: UUID) -> list[OrgMember]:
         """Get all organizations for a user."""
         with session_maker() as session:
             return session.query(OrgMember).filter(OrgMember.user_id == user_id).all()
@@ -68,7 +81,7 @@ class OrgMemberStore:
 
     @staticmethod
     def update_user_role_in_org(
-        org_id: UUID, user_id: int, role_id: int, status: Optional[str] = None
+        org_id: UUID, user_id: UUID, role_id: int, status: Optional[str] = None
     ) -> Optional[OrgMember]:
         """Update user's role in an organization."""
         with session_maker() as session:
@@ -90,7 +103,7 @@ class OrgMemberStore:
             return org_member
 
     @staticmethod
-    def remove_user_from_org(org_id: UUID, user_id: int) -> bool:
+    def remove_user_from_org(org_id: UUID, user_id: UUID) -> bool:
         """Remove a user from an organization."""
         with session_maker() as session:
             org_member = (
@@ -123,3 +136,36 @@ class OrgMemberStore:
             if (normalized := c.name.lstrip('_')) and hasattr(user_settings, normalized)
         }
         return kwargs
+
+    @staticmethod
+    async def get_org_members_paginated(
+        org_id: UUID,
+        offset: int = 0,
+        limit: int = 100,
+    ) -> tuple[list[OrgMember], bool]:
+        """Get paginated list of organization members with user and role info.
+
+        Returns:
+            Tuple of (members_list, has_more) where has_more indicates if there are more results.
+        """
+        async with a_session_maker() as session:
+            # Query for limit + 1 items to determine if there are more results
+            # Order by user_id for consistent pagination
+            query = (
+                select(OrgMember)
+                .options(joinedload(OrgMember.user), joinedload(OrgMember.role))
+                .filter(OrgMember.org_id == org_id)
+                .order_by(OrgMember.user_id)
+                .offset(offset)
+                .limit(limit + 1)
+            )
+            result = await session.execute(query)
+            members = list(result.scalars().all())
+
+            # Check if there are more results
+            has_more = len(members) > limit
+            if has_more:
+                # Remove the extra item
+                members = members[:limit]
+
+            return members, has_more

enterprise/storage/org_service.py

@@ -9,8 +9,11 @@ from uuid import UUID as parse_uuid
 from server.constants import ORG_SETTINGS_VERSION, get_default_litellm_model
 from server.routes.org_models import (
     LiteLLMIntegrationError,
+    OrgAuthorizationError,
     OrgDatabaseError,
     OrgNameExistsError,
+    OrgNotFoundError,
+    OrgUpdate,
 )
 from storage.lite_llm_manager import LiteLlmManager
 from storage.org import Org
@@ -393,6 +396,243 @@ class OrgService:
             )
             return e
 
+    @staticmethod
+    def has_admin_or_owner_role(user_id: str, org_id: UUID) -> bool:
+        """
+        Check if user has admin or owner role in the specified organization.
+
+        Args:
+            user_id: User ID to check
+            org_id: Organization ID to check membership in
+
+        Returns:
+            bool: True if user has admin or owner role, False otherwise
+        """
+        try:
+            # Parse user_id as UUID for database query
+            user_uuid = parse_uuid(user_id)
+
+            # Get the user's membership in this organization
+            # Note: The type annotation says int but the actual column is UUID
+            org_member = OrgMemberStore.get_org_member(org_id, user_uuid)
+            if not org_member:
+                return False
+
+            # Get the role details
+            role = RoleStore.get_role_by_id(org_member.role_id)
+            if not role:
+                return False
+
+            # Admin and owner roles have elevated permissions
+            # Based on test files, both admin and owner have rank 1
+            return role.name in ['admin', 'owner']
+
+        except Exception as e:
+            logger.warning(
+                'Error checking user role in organization',
+                extra={
+                    'user_id': user_id,
+                    'org_id': str(org_id),
+                    'error': str(e),
+                },
+            )
+            return False
+
+    @staticmethod
+    def is_org_member(user_id: str, org_id: UUID) -> bool:
+        """
+        Check if user is a member of the specified organization.
+
+        Args:
+            user_id: User ID to check
+            org_id: Organization ID to check membership in
+
+        Returns:
+            bool: True if user is a member, False otherwise
+        """
+        try:
+            user_uuid = parse_uuid(user_id)
+            org_member = OrgMemberStore.get_org_member(org_id, user_uuid)
+            return org_member is not None
+        except Exception as e:
+            logger.warning(
+                'Error checking user membership in organization',
+                extra={
+                    'user_id': user_id,
+                    'org_id': str(org_id),
+                    'error': str(e),
+                },
+            )
+            return False
+
+    @staticmethod
+    def _get_llm_settings_fields() -> set[str]:
+        """
+        Get the set of organization fields that are considered LLM settings
+        and require admin/owner role to update.
+
+        Returns:
+            set[str]: Set of field names that require elevated permissions
+        """
+        return {
+            'default_llm_model',
+            'default_llm_api_key_for_byor',
+            'default_llm_base_url',
+            'search_api_key',
+            'security_analyzer',
+            'agent',
+            'confirmation_mode',
+            'enable_default_condenser',
+            'condenser_max_size',
+        }
+
+    @staticmethod
+    def _has_llm_settings_updates(update_data: OrgUpdate) -> set[str]:
+        """
+        Check if the update contains any LLM settings fields.
+
+        Args:
+            update_data: The organization update data
+
+        Returns:
+            set[str]: Set of LLM fields being updated (empty if none)
+        """
+        llm_fields = OrgService._get_llm_settings_fields()
+        update_dict = update_data.model_dump(exclude_none=True)
+        return llm_fields.intersection(update_dict.keys())
+
+    @staticmethod
+    async def update_org_with_permissions(
+        org_id: UUID,
+        update_data: OrgUpdate,
+        user_id: str,
+    ) -> Org:
+        """
+        Update organization with permission checks for LLM settings.
+
+        Args:
+            org_id: Organization UUID to update
+            update_data: Organization update data from request
+            user_id: ID of the user requesting the update
+
+        Returns:
+            Org: The updated organization object
+
+        Raises:
+            ValueError: If organization not found
+            PermissionError: If user is not a member, or lacks admin/owner role for LLM settings
+            OrgNameExistsError: If new name already exists for another organization
+            OrgDatabaseError: If database update fails
+        """
+        logger.info(
+            'Updating organization with permission checks',
+            extra={
+                'org_id': str(org_id),
+                'user_id': user_id,
+                'has_update_data': update_data is not None,
+            },
+        )
+
+        # Validate organization exists
+        existing_org = OrgStore.get_org_by_id(org_id)
+        if not existing_org:
+            raise ValueError(f'Organization with ID {org_id} not found')
+
+        # Check if user is a member of this organization
+        if not OrgService.is_org_member(user_id, org_id):
+            logger.warning(
+                'Non-member attempted to update organization',
+                extra={
+                    'user_id': user_id,
+                    'org_id': str(org_id),
+                },
+            )
+            raise PermissionError(
+                'User must be a member of the organization to update it'
+            )
+
+        # Check if name is being updated and validate uniqueness
+        if update_data.name is not None:
+            # Check if new name conflicts with another org
+            existing_org_with_name = OrgStore.get_org_by_name(update_data.name)
+            if (
+                existing_org_with_name is not None
+                and existing_org_with_name.id != org_id
+            ):
+                logger.warning(
+                    'Attempted to update organization with duplicate name',
+                    extra={
+                        'user_id': user_id,
+                        'org_id': str(org_id),
+                        'attempted_name': update_data.name,
+                    },
+                )
+                raise OrgNameExistsError(update_data.name)
+
+        # Check if update contains any LLM settings
+        llm_fields_being_updated = OrgService._has_llm_settings_updates(update_data)
+        if llm_fields_being_updated:
+            # Verify user has admin or owner role
+            has_permission = OrgService.has_admin_or_owner_role(user_id, org_id)
+            if not has_permission:
+                logger.warning(
+                    'User attempted to update LLM settings without permission',
+                    extra={
+                        'user_id': user_id,
+                        'org_id': str(org_id),
+                        'attempted_fields': list(llm_fields_being_updated),
+                    },
+                )
+                raise PermissionError(
+                    'Admin or owner role required to update LLM settings'
+                )
+
+            logger.debug(
+                'User has permission to update LLM settings',
+                extra={
+                    'user_id': user_id,
+                    'org_id': str(org_id),
+                    'llm_fields': list(llm_fields_being_updated),
+                },
+            )
+
+        # Convert to dict for OrgStore (excluding None values)
+        update_dict = update_data.model_dump(exclude_none=True)
+        if not update_dict:
+            logger.info(
+                'No fields to update',
+                extra={'org_id': str(org_id), 'user_id': user_id},
+            )
+            return existing_org
+
+        # Perform the update
+        try:
+            updated_org = OrgStore.update_org(org_id, update_dict)
+            if not updated_org:
+                raise OrgDatabaseError('Failed to update organization in database')
+
+            logger.info(
+                'Organization updated successfully',
+                extra={
+                    'org_id': str(org_id),
+                    'user_id': user_id,
+                    'updated_fields': list(update_dict.keys()),
+                },
+            )
+
+            return updated_org
+
+        except Exception as e:
+            logger.error(
+                'Failed to update organization',
+                extra={
+                    'org_id': str(org_id),
+                    'user_id': user_id,
+                    'error': str(e),
+                },
+            )
+            raise OrgDatabaseError(f'Failed to update organization: {str(e)}')
+
     @staticmethod
     async def get_org_credits(user_id: str, org_id: UUID) -> float | None:
         """
@@ -441,3 +681,274 @@ class OrgService:
                 extra={'user_id': user_id, 'org_id': str(org_id), 'error': str(e)},
             )
             return None
+
+    @staticmethod
+    def get_user_orgs_paginated(
+        user_id: str, page_id: str | None = None, limit: int = 100
+    ):
+        """
+        Get paginated list of organizations for a user.
+
+        Args:
+            user_id: User ID (string that will be converted to UUID)
+            page_id: Optional page ID (offset as string) for pagination
+            limit: Maximum number of organizations to return
+
+        Returns:
+            Tuple of (list of Org objects, next_page_id or None)
+        """
+        logger.debug(
+            'Fetching paginated organizations for user',
+            extra={'user_id': user_id, 'page_id': page_id, 'limit': limit},
+        )
+
+        # Convert user_id string to UUID
+        user_uuid = parse_uuid(user_id)
+
+        # Fetch organizations from store
+        orgs, next_page_id = OrgStore.get_user_orgs_paginated(
+            user_id=user_uuid, page_id=page_id, limit=limit
+        )
+
+        logger.debug(
+            'Retrieved organizations for user',
+            extra={
+                'user_id': user_id,
+                'org_count': len(orgs),
+                'has_more': next_page_id is not None,
+            },
+        )
+
+        return orgs, next_page_id
+
+    @staticmethod
+    async def get_org_by_id(org_id: UUID, user_id: str) -> Org:
+        """
+        Get organization by ID with membership validation.
+
+        This method verifies that the user is a member of the organization
+        before returning the organization details.
+
+        Args:
+            org_id: Organization ID
+            user_id: User ID (string that will be converted to UUID)
+
+        Returns:
+            Org: The organization object
+
+        Raises:
+            OrgNotFoundError: If organization not found or user is not a member
+        """
+        logger.info(
+            'Retrieving organization',
+            extra={'user_id': user_id, 'org_id': str(org_id)},
+        )
+
+        # Verify user is a member of the organization
+        org_member = OrgMemberStore.get_org_member(org_id, parse_uuid(user_id))
+        if not org_member:
+            logger.warning(
+                'User is not a member of organization or organization does not exist',
+                extra={'user_id': user_id, 'org_id': str(org_id)},
+            )
+            raise OrgNotFoundError(str(org_id))
+
+        # Retrieve organization
+        org = OrgStore.get_org_by_id(org_id)
+        if not org:
+            logger.error(
+                'Organization not found despite valid membership',
+                extra={'user_id': user_id, 'org_id': str(org_id)},
+            )
+            raise OrgNotFoundError(str(org_id))
+
+        logger.info(
+            'Successfully retrieved organization',
+            extra={
+                'org_id': str(org.id),
+                'org_name': org.name,
+                'user_id': user_id,
+            },
+        )
+
+        return org
+
+    @staticmethod
+    def verify_owner_authorization(user_id: str, org_id: UUID) -> None:
+        """
+        Verify that the user is the owner of the organization.
+
+        Args:
+            user_id: User ID to check
+            org_id: Organization ID
+
+        Raises:
+            OrgNotFoundError: If organization doesn't exist
+            OrgAuthorizationError: If user is not authorized to delete
+        """
+        # Check if organization exists
+        org = OrgStore.get_org_by_id(org_id)
+        if not org:
+            raise OrgNotFoundError(str(org_id))
+
+        # Check if user is a member of the organization
+        org_member = OrgMemberStore.get_org_member(org_id, parse_uuid(user_id))
+        if not org_member:
+            raise OrgAuthorizationError('User is not a member of this organization')
+
+        # Check if user has owner role
+        role = RoleStore.get_role_by_id(org_member.role_id)
+        if not role or role.name != 'owner':
+            raise OrgAuthorizationError(
+                'Only organization owners can delete organizations'
+            )
+
+        logger.debug(
+            'User authorization verified for organization deletion',
+            extra={'user_id': user_id, 'org_id': str(org_id), 'role': role.name},
+        )
+
+    @staticmethod
+    async def delete_org_with_cleanup(user_id: str, org_id: UUID) -> Org:
+        """
+        Delete organization with complete cleanup of all associated data.
+
+        This method performs the complete organization deletion workflow:
+        1. Verifies user authorization (owner only)
+        2. Performs database cascade deletion and LiteLLM cleanup in single transaction
+
+        Args:
+            user_id: User ID requesting deletion (must be owner)
+            org_id: Organization ID to delete
+
+        Returns:
+            Org: The deleted organization details
+
+        Raises:
+            OrgNotFoundError: If organization doesn't exist
+            OrgAuthorizationError: If user is not authorized to delete
+            OrgDatabaseError: If database operations or LiteLLM cleanup fail
+        """
+        logger.info(
+            'Starting organization deletion',
+            extra={'user_id': user_id, 'org_id': str(org_id)},
+        )
+
+        # Step 1: Verify user authorization
+        OrgService.verify_owner_authorization(user_id, org_id)
+
+        # Step 2: Perform database cascade deletion with LiteLLM cleanup in transaction
+        try:
+            deleted_org = await OrgStore.delete_org_cascade(org_id)
+            if not deleted_org:
+                # This shouldn't happen since we verified existence above
+                raise OrgDatabaseError('Organization not found during deletion')
+
+            logger.info(
+                'Organization deletion completed successfully',
+                extra={
+                    'user_id': user_id,
+                    'org_id': str(org_id),
+                    'org_name': deleted_org.name,
+                },
+            )
+
+            return deleted_org
+
+        except Exception as e:
+            logger.error(
+                'Organization deletion failed',
+                extra={'user_id': user_id, 'org_id': str(org_id), 'error': str(e)},
+            )
+            raise OrgDatabaseError(f'Failed to delete organization: {str(e)}')
+
+    @staticmethod
+    async def check_byor_export_enabled(user_id: str) -> bool:
+        """Check if BYOR export is enabled for the user's current org.
+
+        Returns True if the user's current org has byor_export_enabled set to True.
+        Returns False if the user is not found, has no current org, or the flag is False.
+
+        Args:
+            user_id: User ID to check
+
+        Returns:
+            bool: True if BYOR export is enabled, False otherwise
+        """
+        user = await UserStore.get_user_by_id_async(user_id)
+        if not user or not user.current_org_id:
+            return False
+
+        org = OrgStore.get_org_by_id(user.current_org_id)
+        if not org:
+            return False
+
+        return org.byor_export_enabled
+
+    @staticmethod
+    async def switch_org(user_id: str, org_id: UUID) -> Org:
+        """
+        Switch user's current organization to the specified organization.
+
+        This method:
+        1. Validates that the organization exists
+        2. Validates that the user is a member of the organization
+        3. Updates the user's current_org_id
+
+        Args:
+            user_id: User ID (string that will be converted to UUID)
+            org_id: Organization ID to switch to
+
+        Returns:
+            Org: The organization that was switched to
+
+        Raises:
+            OrgNotFoundError: If organization doesn't exist
+            OrgAuthorizationError: If user is not a member of the organization
+            OrgDatabaseError: If database update fails
+        """
+        logger.info(
+            'Switching user organization',
+            extra={'user_id': user_id, 'org_id': str(org_id)},
+        )
+
+        # Step 1: Check if organization exists
+        org = OrgStore.get_org_by_id(org_id)
+        if not org:
+            raise OrgNotFoundError(str(org_id))
+
+        # Step 2: Validate user is a member of the organization
+        if not OrgService.is_org_member(user_id, org_id):
+            logger.warning(
+                'User attempted to switch to organization they are not a member of',
+                extra={'user_id': user_id, 'org_id': str(org_id)},
+            )
+            raise OrgAuthorizationError(
+                'User must be a member of the organization to switch to it'
+            )
+
+        # Step 3: Update user's current_org_id
+        try:
+            updated_user = UserStore.update_current_org(user_id, org_id)
+            if not updated_user:
+                raise OrgDatabaseError('User not found')
+
+            logger.info(
+                'Successfully switched user organization',
+                extra={
+                    'user_id': user_id,
+                    'org_id': str(org_id),
+                    'org_name': org.name,
+                },
+            )
+
+            return org
+
+        except OrgDatabaseError:
+            raise
+        except Exception as e:
+            logger.error(
+                'Failed to switch user organization',
+                extra={'user_id': user_id, 'org_id': str(org_id), 'error': str(e)},
+            )
+            raise OrgDatabaseError(f'Failed to switch organization: {str(e)}')

enterprise/storage/org_store.py

@@ -10,8 +10,11 @@ from server.constants import (
     ORG_SETTINGS_VERSION,
     get_default_litellm_model,
 )
+from server.routes.org_models import OrphanedUserError
+from sqlalchemy import text
 from sqlalchemy.orm import joinedload
 from storage.database import session_maker
+from storage.lite_llm_manager import LiteLlmManager
 from storage.org import Org
 from storage.org_member import OrgMember
 from storage.user import User
@@ -96,6 +99,63 @@ class OrgStore:
             orgs = session.query(Org).all()
             return orgs
 
+    @staticmethod
+    def get_user_orgs_paginated(
+        user_id: UUID, page_id: str | None = None, limit: int = 100
+    ) -> tuple[list[Org], str | None]:
+        """
+        Get paginated list of organizations for a user.
+
+        Args:
+            user_id: User UUID
+            page_id: Optional page ID (offset as string) for pagination
+            limit: Maximum number of organizations to return
+
+        Returns:
+            Tuple of (list of Org objects, next_page_id or None)
+        """
+        with session_maker() as session:
+            # Build query joining OrgMember with Org
+            query = (
+                session.query(Org)
+                .join(OrgMember, Org.id == OrgMember.org_id)
+                .filter(OrgMember.user_id == user_id)
+                .order_by(Org.name)
+            )
+
+            # Apply pagination offset
+            if page_id is not None:
+                try:
+                    offset = int(page_id)
+                    query = query.offset(offset)
+                except ValueError:
+                    # If page_id is not a valid integer, start from beginning
+                    offset = 0
+            else:
+                offset = 0
+
+            # Fetch limit + 1 to check if there are more results
+            query = query.limit(limit + 1)
+            orgs = query.all()
+
+            # Check if there are more results
+            has_more = len(orgs) > limit
+            if has_more:
+                orgs = orgs[:limit]
+
+            # Calculate next page ID
+            next_page_id = None
+            if has_more:
+                next_page_id = str(offset + limit)
+
+            # Validate org versions
+            validated_orgs = [
+                OrgStore._validate_org_version(org) for org in orgs if org
+            ]
+            validated_orgs = [org for org in validated_orgs if org is not None]
+
+            return validated_orgs, next_page_id
+
     @staticmethod
     def update_org(
         org_id: UUID,
@@ -186,3 +246,143 @@ class OrgStore:
             session.commit()
             session.refresh(org)
             return org
+
+    @staticmethod
+    async def delete_org_cascade(org_id: UUID) -> Org | None:
+        """
+        Delete organization and all associated data in cascade, including external LiteLLM cleanup.
+
+        Args:
+            org_id: UUID of the organization to delete
+
+        Returns:
+            Org: The deleted organization object, or None if not found
+
+        Raises:
+            Exception: If database operations or LiteLLM cleanup fail
+        """
+        with session_maker() as session:
+            # First get the organization to return it
+            org = session.query(Org).filter(Org.id == org_id).first()
+            if not org:
+                return None
+
+            try:
+                # 1. Delete conversation data for organization conversations
+                session.execute(
+                    text("""
+                    DELETE FROM conversation_metadata
+                    WHERE conversation_id IN (
+                        SELECT conversation_id FROM conversation_metadata_saas WHERE org_id = :org_id
+                    )
+                    """),
+                    {'org_id': str(org_id)},
+                )
+
+                session.execute(
+                    text("""
+                    DELETE FROM app_conversation_start_task
+                    WHERE app_conversation_id::text IN (
+                        SELECT conversation_id FROM conversation_metadata_saas WHERE org_id = :org_id
+                    )
+                    """),
+                    {'org_id': str(org_id)},
+                )
+
+                # 2. Delete organization-owned data tables (direct org_id foreign keys)
+                session.execute(
+                    text('DELETE FROM billing_sessions WHERE org_id = :org_id'),
+                    {'org_id': str(org_id)},
+                )
+                session.execute(
+                    text(
+                        'DELETE FROM conversation_metadata_saas WHERE org_id = :org_id'
+                    ),
+                    {'org_id': str(org_id)},
+                )
+                session.execute(
+                    text('DELETE FROM custom_secrets WHERE org_id = :org_id'),
+                    {'org_id': str(org_id)},
+                )
+                session.execute(
+                    text('DELETE FROM api_keys WHERE org_id = :org_id'),
+                    {'org_id': str(org_id)},
+                )
+                session.execute(
+                    text('DELETE FROM slack_conversation WHERE org_id = :org_id'),
+                    {'org_id': str(org_id)},
+                )
+                session.execute(
+                    text('DELETE FROM slack_users WHERE org_id = :org_id'),
+                    {'org_id': str(org_id)},
+                )
+                session.execute(
+                    text('DELETE FROM stripe_customers WHERE org_id = :org_id'),
+                    {'org_id': str(org_id)},
+                )
+
+                # 3. Handle users with this as current_org_id BEFORE deleting memberships
+                # Single query to find orphaned users (those with no alternative org)
+                orphaned_users = session.execute(
+                    text("""
+                        SELECT u.id
+                        FROM "user" u
+                        WHERE u.current_org_id = :org_id
+                        AND NOT EXISTS (
+                            SELECT 1 FROM org_member om
+                            WHERE om.user_id = u.id AND om.org_id != :org_id
+                        )
+                    """),
+                    {'org_id': str(org_id)},
+                ).fetchall()
+
+                if orphaned_users:
+                    raise OrphanedUserError([str(row[0]) for row in orphaned_users])
+
+                # Batch update: reassign current_org_id to an alternative org for all affected users
+                session.execute(
+                    text("""
+                        UPDATE "user" u
+                        SET current_org_id = (
+                            SELECT om.org_id FROM org_member om
+                            WHERE om.user_id = u.id AND om.org_id != :org_id
+                            LIMIT 1
+                        )
+                        WHERE u.current_org_id = :org_id
+                    """),
+                    {'org_id': str(org_id)},
+                )
+
+                # 4. Delete organization memberships (now safe)
+                session.execute(
+                    text('DELETE FROM org_member WHERE org_id = :org_id'),
+                    {'org_id': str(org_id)},
+                )
+
+                # 5. Finally delete the organization
+                session.delete(org)
+
+                # 6. Clean up LiteLLM team before committing transaction
+                logger.info(
+                    'Deleting LiteLLM team within database transaction',
+                    extra={'org_id': str(org_id)},
+                )
+                await LiteLlmManager.delete_team(str(org_id))
+
+                # 7. Commit all changes only if everything succeeded
+                session.commit()
+
+                logger.info(
+                    'Successfully deleted organization and all associated data including LiteLLM team',
+                    extra={'org_id': str(org_id), 'org_name': org.name},
+                )
+
+                return org
+
+            except Exception as e:
+                session.rollback()
+                logger.error(
+                    'Failed to delete organization - transaction rolled back',
+                    extra={'org_id': str(org_id), 'error': str(e)},
+                )
+                raise

enterprise/storage/role_store.py

@@ -4,7 +4,9 @@ Store class for managing roles.
 
 from typing import List, Optional
 
-from storage.database import session_maker
+from sqlalchemy import select
+from sqlalchemy.ext.asyncio import AsyncSession
+from storage.database import a_session_maker, session_maker
 from storage.role import Role
 
 
@@ -33,6 +35,20 @@ class RoleStore:
         with session_maker() as session:
             return session.query(Role).filter(Role.name == name).first()
 
+    @staticmethod
+    async def get_role_by_name_async(
+        name: str,
+        session: Optional[AsyncSession] = None,
+    ) -> Optional[Role]:
+        """Get role by name."""
+        if session is not None:
+            result = await session.execute(select(Role).where(Role.name == name))
+            return result.scalars().first()
+
+        async with a_session_maker() as session:
+            result = await session.execute(select(Role).where(Role.name == name))
+            return result.scalars().first()
+
     @staticmethod
     def list_roles() -> List[Role]:
         """List all roles."""

enterprise/storage/saas_conversation_store.py

@@ -34,11 +34,10 @@ class SaasConversationStore(ConversationStore):
     session_maker: sessionmaker
     org_id: UUID | None = None  # will be fetched automatically
 
-    def __init__(self, user_id: str, session_maker: sessionmaker):
+    def __init__(self, user_id: str, org_id: UUID, session_maker: sessionmaker):
         self.user_id = user_id
+        self.org_id = org_id
         self.session_maker = session_maker
-        user = UserStore.get_user_by_id(user_id)
-        self.org_id = user.current_org_id if user else None
 
     def _select_by_id(self, session, conversation_id: str):
         # Join StoredConversationMetadata with ConversationMetadataSaas to filter by user/org
@@ -235,4 +234,6 @@ class SaasConversationStore(ConversationStore):
         cls, config: OpenHandsConfig, user_id: str | None
     ) -> ConversationStore:
         # user_id should not be None in SaaS, should we raise?
-        return SaasConversationStore(str(user_id), session_maker)
+        user = await UserStore.get_user_by_id_async(user_id)
+        org_id = user.current_org_id if user else None
+        return SaasConversationStore(str(user_id), org_id, session_maker)

enterprise/storage/saas_settings_store.py

@@ -8,10 +8,11 @@ from dataclasses import dataclass
 
 from cryptography.fernet import Fernet
 from pydantic import SecretStr
+from server.constants import LITE_LLM_API_URL
 from server.logger import logger
 from sqlalchemy.orm import joinedload, sessionmaker
 from storage.database import session_maker
-from storage.lite_llm_manager import LiteLlmManager
+from storage.lite_llm_manager import LiteLlmManager, get_openhands_cloud_key_alias
 from storage.org import Org
 from storage.org_member import OrgMember
 from storage.org_store import OrgStore
@@ -23,6 +24,7 @@ from openhands.core.config.openhands_config import OpenHandsConfig
 from openhands.server.settings import Settings
 from openhands.storage.settings.settings_store import SettingsStore
 from openhands.utils.async_utils import call_sync_from_async
+from openhands.utils.llm import is_openhands_model
 
 
 @dataclass
@@ -123,7 +125,6 @@ class SaasSettingsStore(SettingsStore):
         with self.session_maker() as session:
             if not item:
                 return None
-            kwargs = item.model_dump(context={'expose_secrets': True})
             user = (
                 session.query(User)
                 .options(joinedload(User.org_members))
@@ -144,23 +145,29 @@ class SaasSettingsStore(SettingsStore):
                     return None
 
             org_id = user.current_org_id
-            # Check if provider is OpenHands and generate API key if needed
-            if self._is_openhands_provider(item):
-                await self._ensure_openhands_api_key(item, str(org_id))
-            org_member = None
+
+            org_member: OrgMember = None
             for om in user.org_members:
                 if om.org_id == org_id:
                     org_member = om
                     break
             if not org_member or not org_member.llm_api_key:
                 return None
-            org = session.query(Org).filter(Org.id == org_id).first()
+
+            org: Org = session.query(Org).filter(Org.id == org_id).first()
             if not org:
                 logger.error(
                     f'Org not found for ID {org_id} as the current org for user {self.user_id}'
                 )
                 return None
 
+            # Check if we need to generate an LLM key.
+            if item.llm_base_url == LITE_LLM_API_URL:
+                await self._ensure_api_key(
+                    item, str(org_id), openhands_type=is_openhands_model(item.llm_model)
+                )
+
+            kwargs = item.model_dump(context={'expose_secrets': True})
             for model in (user, org, org_member):
                 for key, value in kwargs.items():
                     if hasattr(model, key):
@@ -223,32 +230,49 @@ class SaasSettingsStore(SettingsStore):
         fernet_key = b64encode(hashlib.sha256(jwt_secret.encode()).digest())
         return Fernet(fernet_key)
 
-    def _is_openhands_provider(self, item: Settings) -> bool:
-        """Check if the settings use the OpenHands provider."""
-        return bool(item.llm_model and item.llm_model.startswith('openhands/'))
-
-    async def _ensure_openhands_api_key(self, item: Settings, org_id: str) -> None:
+    async def _ensure_api_key(
+        self, item: Settings, org_id: str, openhands_type: bool = False
+    ) -> None:
         """Generate and set the OpenHands API key for the given settings.
 
-        First checks if an existing key with the OpenHands alias exists,
-        and reuses it if found. Otherwise, generates a new key.
+        First checks if an existing key exists for the user and verifies it
+        is valid in LiteLLM. If valid, reuses it. Otherwise, generates a new key.
         """
-        # Generate new key if none exists
-        generated_key = await LiteLlmManager.generate_key(
+
+        # First, check if our current key is valid
+        if item.llm_api_key and not await LiteLlmManager.verify_existing_key(
+            item.llm_api_key.get_secret_value(),
             self.user_id,
             org_id,
-            None,
-            {'type': 'openhands'},
-        )
+            openhands_type=openhands_type,
+        ):
+            generated_key = None
+            if openhands_type:
+                generated_key = await LiteLlmManager.generate_key(
+                    self.user_id,
+                    org_id,
+                    None,
+                    {'type': 'openhands'},
+                )
+            else:
+                # Must delete any existing key with the same alias first
+                key_alias = get_openhands_cloud_key_alias(self.user_id, org_id)
+                await LiteLlmManager.delete_key_by_alias(key_alias=key_alias)
+                generated_key = await LiteLlmManager.generate_key(
+                    self.user_id,
+                    org_id,
+                    key_alias,
+                    None,
+                )
 
-        if generated_key:
-            item.llm_api_key = SecretStr(generated_key)
-            logger.info(
-                'saas_settings_store:store:generated_openhands_key',
-                extra={'user_id': self.user_id},
-            )
-        else:
-            logger.warning(
-                'saas_settings_store:store:failed_to_generate_openhands_key',
-                extra={'user_id': self.user_id},
-            )
+            if generated_key:
+                item.llm_api_key = SecretStr(generated_key)
+                logger.info(
+                    'saas_settings_store:store:generated_openhands_key',
+                    extra={'user_id': self.user_id},
+                )
+            else:
+                logger.warning(
+                    'saas_settings_store:store:failed_to_generate_openhands_key',
+                    extra={'user_id': self.user_id},
+                )

enterprise/storage/user.py

@@ -31,6 +31,8 @@ class User(Base):  # type: ignore
     user_consents_to_analytics = Column(Boolean, nullable=True)
     email = Column(String, nullable=True)
     email_verified = Column(Boolean, nullable=True)
+    git_user_name = Column(String, nullable=True)
+    git_user_email = Column(String, nullable=True)
 
     # Relationships
     role = relationship('Role', back_populates='users')

enterprise/storage/user_store.py

@@ -5,6 +5,7 @@ Store class for managing users.
 import asyncio
 import uuid
 from typing import Optional
+from uuid import UUID
 
 from server.auth.token_manager import TokenManager
 from server.constants import (
@@ -14,15 +15,20 @@ from server.constants import (
     get_default_litellm_model,
 )
 from server.logger import logger
-from sqlalchemy import text
+from sqlalchemy import select, text
 from sqlalchemy.orm import joinedload
-from storage.database import session_maker
-from storage.encrypt_utils import decrypt_legacy_model
+from storage.database import a_session_maker, session_maker
+from storage.encrypt_utils import (
+    decrypt_legacy_model,
+    decrypt_legacy_value,
+    encrypt_legacy_value,
+)
 from storage.org import Org
 from storage.org_member import OrgMember
 from storage.role_store import RoleStore
 from storage.user import User
 from storage.user_settings import UserSettings
+from utils.identity import resolve_display_name
 
 from openhands.utils.async_utils import GENERAL_TIMEOUT, call_async_from_sync
 
@@ -49,7 +55,8 @@ class UserStore:
             org = Org(
                 id=uuid.UUID(user_id),
                 name=f'user_{user_id}_org',
-                contact_name=user_info['preferred_username'],
+                contact_name=resolve_display_name(user_info)
+                or user_info.get('preferred_username', ''),
                 contact_email=user_info['email'],
                 v1_enabled=True,
             )
@@ -116,7 +123,7 @@ class UserStore:
         redis_client = UserStore._get_redis_client()
         if redis_client is None:
             logger.warning(
-                'saas_settings_store:_acquire_user_creation_lock:no_redis_client',
+                'user_store:_acquire_user_creation_lock:no_redis_client',
                 extra={'user_id': user_id},
             )
             return True  # Proceed without locking if Redis is unavailable
@@ -127,6 +134,25 @@ class UserStore:
         )
         return bool(lock_acquired)
 
+    @staticmethod
+    async def _release_user_creation_lock(user_id: str) -> bool:
+        """Release the distributed lock for user creation.
+
+        Returns True if the lock was released or if Redis is unavailable.
+        Returns False if the lock could not be released.
+        """
+        redis_client = UserStore._get_redis_client()
+        if redis_client is None:
+            logger.warning(
+                'user_store:_release_user_creation_lock:no_redis_client',
+                extra={'user_id': user_id},
+            )
+            return True  # Nothing to release if Redis is unavailable
+
+        user_key = f'{_REDIS_USER_CREATION_KEY_PREFIX}{user_id}'
+        deleted = await redis_client.delete(user_key)
+        return bool(deleted)
+
     @staticmethod
     async def migrate_user(
         user_id: str,
@@ -147,24 +173,47 @@ class UserStore:
         )
         decrypted_user_settings = UserSettings(**kwargs)
         with session_maker() as session:
+            # Check if user has completed billing sessions to enable BYOR export
+            from storage.billing_session import BillingSession
+
+            has_completed_billing = (
+                session.query(BillingSession)
+                .filter(
+                    BillingSession.user_id == user_id,
+                    BillingSession.status == 'completed',
+                )
+                .first()
+                is not None
+            )
+
             # create personal org
             org = Org(
                 id=uuid.UUID(user_id),
                 name=f'user_{user_id}_org',
                 org_version=user_settings.user_version,
-                contact_name=user_info['username'],
+                contact_name=resolve_display_name(user_info)
+                or user_info.get('username', ''),
                 contact_email=user_info['email'],
+                byor_export_enabled=has_completed_billing,
             )
             session.add(org)
 
             from storage.lite_llm_manager import LiteLlmManager
 
+            logger.debug(
+                'user_store:migrate_user:calling_litellm_migrate_entries',
+                extra={'user_id': user_id},
+            )
             await LiteLlmManager.migrate_entries(
                 str(org.id),
                 user_id,
                 decrypted_user_settings,
             )
 
+            logger.debug(
+                'user_store:migrate_user:done_litellm_migrate_entries',
+                extra={'user_id': user_id},
+            )
             custom_settings = UserStore._has_custom_settings(
                 decrypted_user_settings, user_settings.user_version
             )
@@ -172,7 +221,15 @@ class UserStore:
             # avoids circular reference. This migrate method is temprorary until all users are migrated.
             from integrations.stripe_service import migrate_customer
 
+            logger.debug(
+                'user_store:migrate_user:calling_stripe_migrate_customer',
+                extra={'user_id': user_id},
+            )
             await migrate_customer(session, user_id, org)
+            logger.debug(
+                'user_store:migrate_user:done_stripe_migrate_customer',
+                extra={'user_id': user_id},
+            )
 
             from storage.org_store import OrgStore
 
@@ -201,7 +258,15 @@ class UserStore:
             )
             session.add(user)
 
-            role = RoleStore.get_role_by_name('owner')
+            logger.debug(
+                'user_store:migrate_user:calling_get_role_by_name',
+                extra={'user_id': user_id},
+            )
+            role = await RoleStore.get_role_by_name_async('owner')
+            logger.debug(
+                'user_store:migrate_user:done_get_role_by_name',
+                extra={'user_id': user_id},
+            )
 
             from storage.org_member_store import OrgMemberStore
 
@@ -214,7 +279,6 @@ class UserStore:
             if not custom_settings:
                 del org_member_kwargs['llm_model']
                 del org_member_kwargs['llm_base_url']
-                del org_member_kwargs['llm_api_key_for_byor']
 
             org_member = OrgMember(
                 org_id=org.id,
@@ -229,6 +293,10 @@ class UserStore:
             user_settings.already_migrated = True
             session.merge(user_settings)
             session.flush()
+            logger.debug(
+                'user_store:migrate_user:session_flush_complete',
+                extra={'user_id': user_id},
+            )
 
             # need to migrate conversation metadata
             session.execute(
@@ -296,8 +364,272 @@ class UserStore:
             session.commit()
             session.refresh(user)
             user.org_members  # load org_members
+            logger.debug(
+                'user_store:migrate_user:session_committed',
+                extra={'user_id': user_id},
+            )
             return user
 
+    @staticmethod
+    async def downgrade_user(user_id: str) -> UserSettings | None:
+        """
+        This method can be removed once orgs is established - probably after Feb 15 2026
+        Downgrade a migrated user back to the pre-migration state.
+
+        This reverses the migrate_user operation:
+        1. Get the user's settings from user_settings table (migrated users) or
+           create new user_settings from org_members table (new sign-ups)
+        2. Call LiteLlmManager.downgrade_entries to revert LiteLLM state
+        3. Copy user_id from conversation_metadata_saas to conversation_metadata
+        4. Delete conversation_metadata_saas entries
+        5. Reset org_id columns in related tables (stripe_customers, slack_users, etc.)
+        6. Delete the org_member and org entries
+        7. Delete the user entry
+        8. Set already_migrated=False on user_settings
+
+        For new sign-ups (users who registered after migration was deployed),
+        there won't be an existing user_settings entry. In this case, we fall back
+        to the org_members table to get the user's API keys and settings, and create
+        a new user_settings entry for them.
+
+        Args:
+            user_id: The Keycloak user ID to downgrade
+
+        Returns:
+            The user_settings if downgrade was successful, None otherwise.
+            Returns None if the org has multiple members (not a personal org).
+        """
+        logger.info(
+            'user_store:downgrade_user:start',
+            extra={'user_id': user_id},
+        )
+
+        with session_maker() as session:
+            # Get the user and their org_member
+            user = (
+                session.query(User)
+                .options(joinedload(User.org_members))
+                .filter(User.id == uuid.UUID(user_id))
+                .first()
+            )
+            if not user:
+                logger.warning(
+                    'user_store:downgrade_user:user_not_found',
+                    extra={'user_id': user_id},
+                )
+                return None
+
+            # Get the user's personal org (org_id == user_id)
+            org = session.query(Org).filter(Org.id == uuid.UUID(user_id)).first()
+            if not org:
+                logger.warning(
+                    'user_store:downgrade_user:org_not_found',
+                    extra={'user_id': user_id},
+                )
+                return None
+
+            # Get org_members for this org - should only be one for personal orgs
+            org_members = (
+                session.query(OrgMember).filter(OrgMember.org_id == org.id).all()
+            )
+
+            if len(org_members) != 1:
+                logger.error(
+                    'user_store:downgrade_user:unexpected_org_members_count',
+                    extra={
+                        'user_id': user_id,
+                        'org_id': str(org.id),
+                        'org_members_count': len(org_members),
+                    },
+                )
+                return None
+
+            org_member = org_members[0]
+
+            # Get the user_settings (for migrated users)
+            user_settings = (
+                session.query(UserSettings)
+                .filter(
+                    UserSettings.keycloak_user_id == user_id,
+                    UserSettings.already_migrated.is_(True),
+                )
+                .first()
+            )
+
+            # For new sign-ups after migration, user_settings won't exist
+            # Fall back to getting data from org_members
+            if user_settings:
+                if org_member.llm_api_key and org_member.llm_api_key.get_secret_value():
+                    user_settings.llm_api_key = encrypt_legacy_value(
+                        org_member.llm_api_key.get_secret_value()
+                    )
+                if (
+                    org_member.llm_api_key_for_byor
+                    and org_member.llm_api_key_for_byor.get_secret_value()
+                ):
+                    user_settings.llm_api_key_for_byor = encrypt_legacy_value(
+                        org_member.llm_api_key_for_byor.get_secret_value()
+                    )
+                logger.info(
+                    'user_store:downgrade_user:updated_user_settings_from_org_member',
+                    extra={'user_id': user_id},
+                )
+            else:
+                # Create a new user_settings entry from OrgMember, User, and Org data
+                # This is needed for new sign-ups who don't have user_settings
+                user_settings = UserStore._create_user_settings_from_entities(
+                    user_id, org_member, user, org
+                )
+                session.add(user_settings)
+                logger.info(
+                    'user_store:downgrade_user:created_user_settings_from_org_member',
+                    extra={'user_id': user_id},
+                )
+            session.flush()
+
+            # Call LiteLLM downgrade
+            from storage.lite_llm_manager import LiteLlmManager
+
+            logger.debug(
+                'user_store:downgrade_user:calling_litellm_downgrade_entries',
+                extra={'user_id': user_id},
+            )
+
+            encrypted_fields = [
+                'llm_api_key',
+                'llm_api_key_for_byor',
+                'search_api_key',
+                'sandbox_api_key',
+            ]
+            for field in encrypted_fields:
+                value = getattr(user_settings, field, None)
+                if value:
+                    try:
+                        value = decrypt_legacy_value(value)
+                        setattr(user_settings, field, value)
+                    except Exception:
+                        pass
+
+            await LiteLlmManager.downgrade_entries(
+                str(org.id),
+                user_id,
+                user_settings,
+            )
+            logger.debug(
+                'user_store:downgrade_user:done_litellm_downgrade_entries',
+                extra={'user_id': user_id},
+            )
+
+            user_uuid = uuid.UUID(user_id)
+
+            # Step 3: Copy user_id from conversation_metadata_saas to conversation_metadata
+            # This ensures any conversations created after migration have their user_id
+            # preserved in the original table before we delete the saas entries
+            session.execute(
+                text("""
+                    UPDATE conversation_metadata
+                    SET user_id = :user_id
+                    WHERE conversation_id IN (
+                        SELECT conversation_id
+                        FROM conversation_metadata_saas
+                        WHERE user_id = :user_uuid
+                    )
+                """),
+                {'user_id': user_id, 'user_uuid': user_uuid},
+            )
+
+            # Step 4: Delete conversation_metadata_saas entries
+            session.execute(
+                text('DELETE FROM conversation_metadata_saas WHERE user_id = :user_id'),
+                {'user_id': user_uuid},
+            )
+
+            # Step 5: Reset org_id columns in related tables
+            # Reset stripe_customers
+            session.execute(
+                text(
+                    'UPDATE stripe_customers SET org_id = NULL WHERE org_id = :org_id'
+                ),
+                {'org_id': user_uuid},
+            )
+
+            # Reset slack_users
+            session.execute(
+                text('UPDATE slack_users SET org_id = NULL WHERE org_id = :org_id'),
+                {'org_id': user_uuid},
+            )
+
+            # Reset slack_conversation
+            session.execute(
+                text(
+                    'UPDATE slack_conversation SET org_id = NULL WHERE org_id = :org_id'
+                ),
+                {'org_id': user_uuid},
+            )
+
+            # Reset api_keys
+            session.execute(
+                text('UPDATE api_keys SET org_id = NULL WHERE org_id = :org_id'),
+                {'org_id': user_uuid},
+            )
+
+            # Reset custom_secrets
+            session.execute(
+                text('UPDATE custom_secrets SET org_id = NULL WHERE org_id = :org_id'),
+                {'org_id': user_uuid},
+            )
+
+            # Reset billing_sessions
+            session.execute(
+                text(
+                    'UPDATE billing_sessions SET org_id = NULL WHERE org_id = :org_id'
+                ),
+                {'org_id': user_uuid},
+            )
+
+            # Step 6: Delete org_member entries for this org
+            session.execute(
+                text('DELETE FROM org_member WHERE org_id = :org_id'),
+                {'org_id': user_uuid},
+            )
+
+            # Step 7: Delete the user entry
+            session.execute(
+                text('DELETE FROM "user" WHERE id = :user_id'),
+                {'user_id': user_uuid},
+            )
+
+            # Delete the org entry
+            session.execute(
+                text('DELETE FROM org WHERE id = :org_id'),
+                {'org_id': user_uuid},
+            )
+
+            # Step 8: Set already_migrated=False on user_settings and encrypt fields
+            user_settings.already_migrated = False
+
+            # Re-encrypt the sensitive fields before storing in the DB
+            encrypt_keys = [
+                'llm_api_key',
+                'llm_api_key_for_byor',
+                'search_api_key',
+                'sandbox_api_key',
+            ]
+            for key in encrypt_keys:
+                value = getattr(user_settings, key, None)
+                if value is not None and not _is_legacy_value_encrypted(value):
+                    setattr(user_settings, key, encrypt_legacy_value(value))
+
+            session.merge(user_settings)
+
+            session.commit()
+
+            logger.info(
+                'user_store:downgrade_user:complete',
+                extra={'user_id': user_id},
+            )
+            return user_settings
+
     @staticmethod
     def get_user_by_id(user_id: str) -> Optional[User]:
         """Get user by Keycloak user ID (sync version).
@@ -322,48 +654,53 @@ class UserStore:
             ):
                 # The user is already being created in another thread / process
                 logger.info(
-                    'saas_settings_store:create_default_settings:waiting_for_lock',
+                    'user_store:create_default_settings:waiting_for_lock',
                     extra={'user_id': user_id},
                 )
                 call_async_from_sync(
                     asyncio.sleep, GENERAL_TIMEOUT, _RETRY_LOAD_DELAY_SECONDS
                 )
 
-            # Check for user again as migration could have happened while trying to get the lock.
-            user = (
-                session.query(User)
-                .options(joinedload(User.org_members))
-                .filter(User.id == uuid.UUID(user_id))
-                .first()
-            )
-            if user:
-                return user
+            try:
+                # Check for user again as migration could have happened while trying to get the lock.
+                user = (
+                    session.query(User)
+                    .options(joinedload(User.org_members))
+                    .filter(User.id == uuid.UUID(user_id))
+                    .first()
+                )
+                if user:
+                    return user
 
-            user_settings = (
-                session.query(UserSettings)
-                .filter(
-                    UserSettings.keycloak_user_id == user_id,
-                    UserSettings.already_migrated.is_(False),
+                user_settings = (
+                    session.query(UserSettings)
+                    .filter(
+                        UserSettings.keycloak_user_id == user_id,
+                        UserSettings.already_migrated.is_(False),
+                    )
+                    .first()
                 )
-                .first()
-            )
-            if user_settings:
-                token_manager = TokenManager()
-                user_info = call_async_from_sync(
-                    token_manager.get_user_info_from_user_id,
-                    GENERAL_TIMEOUT,
-                    user_id,
+                if user_settings:
+                    token_manager = TokenManager()
+                    user_info = call_async_from_sync(
+                        token_manager.get_user_info_from_user_id,
+                        GENERAL_TIMEOUT,
+                        user_id,
+                    )
+                    user = call_async_from_sync(
+                        UserStore.migrate_user,
+                        GENERAL_TIMEOUT,
+                        user_id,
+                        user_settings,
+                        user_info,
+                    )
+                    return user
+                else:
+                    return None
+            finally:
+                call_async_from_sync(
+                    UserStore._release_user_creation_lock, GENERAL_TIMEOUT, user_id
                 )
-                user = call_async_from_sync(
-                    UserStore.migrate_user,
-                    GENERAL_TIMEOUT,
-                    user_id,
-                    user_settings,
-                    user_info,
-                )
-                return user
-            else:
-                return None
 
     @staticmethod
     async def get_user_by_id_async(user_id: str) -> Optional[User]:
@@ -372,13 +709,13 @@ class UserStore:
         This is the preferred method when calling from an async context as it
         avoids event loop conflicts that can occur with the sync version.
         """
-        with session_maker() as session:
-            user = (
-                session.query(User)
+        async with a_session_maker() as session:
+            result = await session.execute(
+                select(User)
                 .options(joinedload(User.org_members))
                 .filter(User.id == uuid.UUID(user_id))
-                .first()
             )
+            user = result.scalars().first()
             if user:
                 return user
 
@@ -386,40 +723,50 @@ class UserStore:
             while not await UserStore._acquire_user_creation_lock(user_id):
                 # The user is already being created in another thread / process
                 logger.info(
-                    'saas_settings_store:create_default_settings:waiting_for_lock',
+                    'user_store:get_user_by_id_async:waiting_for_lock',
                     extra={'user_id': user_id},
                 )
                 await asyncio.sleep(_RETRY_LOAD_DELAY_SECONDS)
 
-            # Check for user again as migration could have happened while trying to get the lock.
-            user = (
-                session.query(User)
-                .options(joinedload(User.org_members))
-                .filter(User.id == uuid.UUID(user_id))
-                .first()
-            )
-            if user:
-                return user
+            try:
+                # Check for user again as migration could have happened while trying to get the lock.
+                result = await session.execute(
+                    select(User)
+                    .options(joinedload(User.org_members))
+                    .filter(User.id == uuid.UUID(user_id))
+                )
+                user = result.scalars().first()
+                if user:
+                    return user
 
-            user_settings = (
-                session.query(UserSettings)
-                .filter(
-                    UserSettings.keycloak_user_id == user_id,
-                    UserSettings.already_migrated.is_(False),
+                logger.info(
+                    'user_store:get_user_by_id_async:start_migration',
+                    extra={'user_id': user_id},
                 )
-                .first()
-            )
-            if user_settings:
-                token_manager = TokenManager()
-                user_info = await token_manager.get_user_info_from_user_id(user_id)
-                user = await UserStore.migrate_user(
-                    user_id,
-                    user_settings,
-                    user_info,
+                result = await session.execute(
+                    select(UserSettings).filter(
+                        UserSettings.keycloak_user_id == user_id,
+                        UserSettings.already_migrated.is_(False),
+                    )
                 )
-                return user
-            else:
-                return None
+                user_settings = result.scalars().first()
+                if user_settings:
+                    token_manager = TokenManager()
+                    user_info = await token_manager.get_user_info_from_user_id(user_id)
+                    logger.info(
+                        'user_store:get_user_by_id_async:calling_migrate_user',
+                        extra={'user_id': user_id},
+                    )
+                    user = await UserStore.migrate_user(
+                        user_id,
+                        user_settings,
+                        user_info,
+                    )
+                    return user
+                else:
+                    return None
+            finally:
+                await UserStore._release_user_creation_lock(user_id)
 
     @staticmethod
     def list_users() -> list[User]:
@@ -427,6 +774,75 @@ class UserStore:
         with session_maker() as session:
             return session.query(User).all()
 
+    @staticmethod
+    def update_current_org(user_id: str, org_id: UUID) -> Optional[User]:
+        """Update the user's current organization.
+
+        Args:
+            user_id: The user's ID (Keycloak user ID)
+            org_id: The organization ID to set as current
+
+        Returns:
+            User: The updated user object, or None if user not found
+        """
+        with session_maker() as session:
+            user = (
+                session.query(User)
+                .filter(User.id == uuid.UUID(user_id))
+                .with_for_update()
+                .first()
+            )
+            if not user:
+                return None
+
+            user.current_org_id = org_id
+            session.commit()
+            session.refresh(user)
+            return user
+
+    @staticmethod
+    async def backfill_contact_name(user_id: str, user_info: dict) -> None:
+        """Update contact_name on the personal org if it still has a username-style value.
+
+        Called during login to gradually fix existing users whose contact_name
+        was stored as their username (before the resolve_display_name fix).
+        Preserves custom values that were set via the PATCH endpoint.
+        """
+        real_name = resolve_display_name(user_info)
+        if not real_name:
+            logger.debug(
+                'backfill_contact_name:no_real_name',
+                extra={'user_id': user_id},
+            )
+            return
+
+        preferred_username = user_info.get('preferred_username', '')
+        username = user_info.get('username', '')
+
+        async with a_session_maker() as session:
+            result = await session.execute(
+                select(Org).filter(Org.id == uuid.UUID(user_id))
+            )
+            org = result.scalars().first()
+            if not org:
+                logger.debug(
+                    'backfill_contact_name:org_not_found',
+                    extra={'user_id': user_id},
+                )
+                return
+
+            if org.contact_name in (preferred_username, username):
+                logger.info(
+                    'backfill_contact_name:updated',
+                    extra={
+                        'user_id': user_id,
+                        'old': org.contact_name,
+                        'new': real_name,
+                    },
+                )
+                org.contact_name = real_name
+                await session.commit()
+
     # Prevent circular imports
     from typing import TYPE_CHECKING
 
@@ -481,6 +897,96 @@ class UserStore:
         }
         return kwargs
 
+    @staticmethod
+    def _create_user_settings_from_entities(
+        user_id: str, org_member: OrgMember, user: User, org: Org
+    ) -> UserSettings:
+        """Create UserSettings from OrgMember, User, and Org data.
+
+        Uses OrgMember values first. If an OrgMember field is None and there's
+        a corresponding "default_" field in Org, use the Org value.
+        Also pulls relevant fields from User.
+
+        Args:
+            user_id: The Keycloak user ID
+            org_member: The OrgMember entity
+            user: The User entity
+            org: The Org entity
+
+        Returns:
+            A new UserSettings object populated from the entities
+        """
+        # Mapping from OrgMember fields to corresponding Org "default_" fields
+        org_member_to_org_default = {
+            'llm_model': 'default_llm_model',
+            'llm_base_url': 'default_llm_base_url',
+            'max_iterations': 'default_max_iterations',
+        }
+
+        def get_value_with_org_fallback(field_name: str, org_member_value):
+            """Get value from OrgMember, falling back to Org default if None."""
+            if org_member_value is not None:
+                return org_member_value
+            org_default_field = org_member_to_org_default.get(field_name)
+            if org_default_field and hasattr(org, org_default_field):
+                return getattr(org, org_default_field)
+            return None
+
+        # Get values from OrgMember with Org fallback for fields with default_ prefix
+        llm_model = get_value_with_org_fallback('llm_model', org_member.llm_model)
+        llm_base_url = get_value_with_org_fallback(
+            'llm_base_url', org_member.llm_base_url
+        )
+        max_iterations = get_value_with_org_fallback(
+            'max_iterations', org_member.max_iterations
+        )
+
+        return UserSettings(
+            keycloak_user_id=user_id,
+            # OrgMember fields
+            llm_api_key=org_member.llm_api_key.get_secret_value()
+            if org_member.llm_api_key
+            else None,
+            llm_api_key_for_byor=org_member.llm_api_key_for_byor.get_secret_value()
+            if org_member.llm_api_key_for_byor
+            else None,
+            llm_model=llm_model,
+            llm_base_url=llm_base_url,
+            max_iterations=max_iterations,
+            # User fields
+            accepted_tos=user.accepted_tos,
+            enable_sound_notifications=user.enable_sound_notifications,
+            language=user.language,
+            user_consents_to_analytics=user.user_consents_to_analytics,
+            email=user.email,
+            email_verified=user.email_verified,
+            git_user_name=user.git_user_name,
+            git_user_email=user.git_user_email,
+            # Org fields
+            agent=org.agent,
+            security_analyzer=org.security_analyzer,
+            confirmation_mode=org.confirmation_mode,
+            remote_runtime_resource_factor=org.remote_runtime_resource_factor,
+            enable_default_condenser=org.enable_default_condenser,
+            billing_margin=org.billing_margin,
+            enable_proactive_conversation_starters=org.enable_proactive_conversation_starters,
+            sandbox_base_container_image=org.sandbox_base_container_image,
+            sandbox_runtime_container_image=org.sandbox_runtime_container_image,
+            user_version=org.org_version,
+            mcp_config=org.mcp_config,
+            search_api_key=org.search_api_key.get_secret_value()
+            if org.search_api_key
+            else None,
+            sandbox_api_key=org.sandbox_api_key.get_secret_value()
+            if org.sandbox_api_key
+            else None,
+            max_budget_per_task=org.max_budget_per_task,
+            enable_solvability_analysis=org.enable_solvability_analysis,
+            v1_enabled=org.v1_enabled,
+            condenser_max_size=org.condenser_max_size,
+            already_migrated=False,
+        )
+
     @staticmethod
     def _has_custom_settings(
         user_settings: UserSettings, old_user_version: int | None
@@ -526,3 +1032,12 @@ class UserStore:
                 return False  # Matches old default
 
         return True  # Custom model
+
+
+def _is_legacy_value_encrypted(value: str) -> bool:
+    """Check if a legacy value is encrypted by trying to decrypt it"""
+    try:
+        decrypt_legacy_value(value)
+        return True
+    except Exception:
+        return False

enterprise/sync/resend_keycloak.py

@@ -26,6 +26,7 @@ Optional environment variables:
 """
 
 import os
+import re
 import sys
 import time
 from typing import Any, Dict, List, Optional
@@ -90,6 +91,31 @@ class ResendAPIError(ResendSyncError):
     pass
 
 
+# Email validation regex pattern - matches standard email format
+# This pattern is intentionally strict to avoid Resend API validation errors
+# It rejects special characters like ! that some email providers technically allow
+# but Resend's API does not accept
+EMAIL_REGEX = re.compile(r'^[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}$')
+
+
+def is_valid_email(email: str) -> bool:
+    """Validate an email address format.
+
+    This uses a regex pattern that matches most valid email addresses
+    while rejecting addresses with special characters that Resend's API
+    does not accept (e.g., exclamation marks).
+
+    Args:
+        email: The email address to validate.
+
+    Returns:
+        True if the email is valid, False otherwise.
+    """
+    if not email:
+        return False
+    return bool(EMAIL_REGEX.match(email))
+
+
 def get_keycloak_users(offset: int = 0, limit: int = 100) -> List[Dict[str, Any]]:
     """Get users from Keycloak using the admin client.
 
@@ -336,6 +362,7 @@ def sync_users_to_resend():
             'total_users': total_users,
             'existing_contacts': len(resend_contacts),
             'added_contacts': 0,
+            'skipped_invalid_emails': 0,
             'errors': 0,
         }
 
@@ -355,6 +382,12 @@ def sync_users_to_resend():
                     logger.debug(f'User {email} already exists in Resend, skipping')
                     continue
 
+                # Validate email format before attempting to add to Resend
+                if not is_valid_email(email):
+                    logger.warning(f'Skipping user with invalid email format: {email}')
+                    stats['skipped_invalid_emails'] += 1
+                    continue
+
                 try:
                     first_name = user.get('first_name')
                     last_name = user.get('last_name')

enterprise/tests/unit/integrations/test_resolver_context.py

@@ -141,12 +141,14 @@ def test_custom_to_static_conversion():
 
 def create_provider_tokens(
     tokens_dict: dict[ProviderType, str],
-) -> dict[ProviderType, ProviderToken]:
-    """Helper to create provider tokens dictionary."""
-    return {
-        provider_type: ProviderToken(token=SecretStr(token_value))
-        for provider_type, token_value in tokens_dict.items()
-    }
+) -> MappingProxyType:
+    """Helper to create provider tokens as MappingProxyType."""
+    return MappingProxyType(
+        {
+            provider_type: ProviderToken(token=SecretStr(token_value))
+            for provider_type, token_value in tokens_dict.items()
+        }
+    )
 
 
 @pytest.mark.asyncio
@@ -264,3 +266,63 @@ async def test_get_latest_token_can_be_used_with_static_secret(
     # Assert - this should NOT raise a ValidationError
     static_secret = StaticSecret(value=token, description='GITHUB authentication token')
     assert static_secret.get_value() == token_value
+
+
+# ---------------------------------------------------------------------------
+# Tests for get_authenticated_git_url - ensuring proper authenticated URLs
+# ---------------------------------------------------------------------------
+
+
+@pytest.mark.asyncio
+async def test_get_authenticated_git_url_raises_when_no_tokens(
+    resolver_context, mock_saas_user_auth
+):
+    """Test that get_authenticated_git_url raises error when no provider tokens available."""
+    # Arrange
+    mock_saas_user_auth.get_provider_tokens = AsyncMock(return_value=None)
+
+    # Act & Assert
+    with pytest.raises(ValueError, match='No provider tokens available'):
+        await resolver_context.get_authenticated_git_url('owner/repo')
+
+
+@pytest.mark.asyncio
+async def test_get_provider_handler_caches_instance(
+    resolver_context, mock_saas_user_auth
+):
+    """Test that _get_provider_handler caches the handler instance."""
+    # Arrange
+    token_value = 'ghp_test_token'
+    provider_tokens = create_provider_tokens({ProviderType.GITHUB: token_value})
+    mock_saas_user_auth.get_provider_tokens = AsyncMock(return_value=provider_tokens)
+    mock_saas_user_auth.get_user_id = AsyncMock(return_value='test-user-id')
+
+    # Act - call _get_provider_handler twice
+    handler1 = await resolver_context._get_provider_handler()
+    handler2 = await resolver_context._get_provider_handler()
+
+    # Assert - should be the same instance (cached)
+    assert handler1 is handler2
+    # get_provider_tokens should only be called once
+    assert mock_saas_user_auth.get_provider_tokens.call_count == 1
+
+
+@pytest.mark.asyncio
+async def test_get_provider_handler_creates_handler_with_correct_params(
+    resolver_context, mock_saas_user_auth
+):
+    """Test that _get_provider_handler creates ProviderHandler with correct parameters."""
+    # Arrange
+    token_value = 'ghp_test_token'
+    provider_tokens = create_provider_tokens({ProviderType.GITHUB: token_value})
+    mock_saas_user_auth.get_provider_tokens = AsyncMock(return_value=provider_tokens)
+    mock_saas_user_auth.get_user_id = AsyncMock(return_value='test-user-id')
+
+    # Act
+    handler = await resolver_context._get_provider_handler()
+
+    # Assert
+    from openhands.integrations.provider import ProviderHandler
+
+    assert isinstance(handler, ProviderHandler)
+    assert handler.provider_tokens == provider_tokens

enterprise/tests/unit/server/routes/test_api_keys.py

@@ -6,6 +6,7 @@ import httpx
 import pytest
 from fastapi import HTTPException
 from server.routes.api_keys import (
+    check_byor_permitted,
     delete_byor_key_from_litellm,
     get_llm_api_key_for_byor,
 )
@@ -182,16 +183,18 @@ class TestGetLlmApiKeyForByor:
     """Test the get_llm_api_key_for_byor endpoint."""
 
     @pytest.mark.asyncio
+    @patch('storage.org_service.OrgService.check_byor_export_enabled')
     @patch('server.routes.api_keys.store_byor_key_in_db')
     @patch('server.routes.api_keys.generate_byor_key')
     @patch('server.routes.api_keys.get_byor_key_from_db')
     async def test_no_key_in_database_generates_new(
-        self, mock_get_key, mock_generate_key, mock_store_key
+        self, mock_get_key, mock_generate_key, mock_store_key, mock_check_enabled
     ):
         """Test that when no key exists in database, a new one is generated."""
         # Arrange
         user_id = 'user-123'
         new_key = 'sk-new-generated-key'
+        mock_check_enabled.return_value = True
         mock_get_key.return_value = None
         mock_generate_key.return_value = new_key
         mock_store_key.return_value = None
@@ -201,20 +204,23 @@ class TestGetLlmApiKeyForByor:
 
         # Assert
         assert result == {'key': new_key}
+        mock_check_enabled.assert_called_once_with(user_id)
         mock_get_key.assert_called_once_with(user_id)
         mock_generate_key.assert_called_once_with(user_id)
         mock_store_key.assert_called_once_with(user_id, new_key)
 
     @pytest.mark.asyncio
+    @patch('storage.org_service.OrgService.check_byor_export_enabled')
     @patch('storage.lite_llm_manager.LiteLlmManager.verify_key')
     @patch('server.routes.api_keys.get_byor_key_from_db')
     async def test_valid_key_in_database_returns_key(
-        self, mock_get_key, mock_verify_key
+        self, mock_get_key, mock_verify_key, mock_check_enabled
     ):
         """Test that when a valid key exists in database, it is returned."""
         # Arrange
         user_id = 'user-123'
         existing_key = 'sk-existing-valid-key'
+        mock_check_enabled.return_value = True
         mock_get_key.return_value = existing_key
         mock_verify_key.return_value = True
 
@@ -223,10 +229,12 @@ class TestGetLlmApiKeyForByor:
 
         # Assert
         assert result == {'key': existing_key}
+        mock_check_enabled.assert_called_once_with(user_id)
         mock_get_key.assert_called_once_with(user_id)
         mock_verify_key.assert_called_once_with(existing_key, user_id)
 
     @pytest.mark.asyncio
+    @patch('storage.org_service.OrgService.check_byor_export_enabled')
     @patch('server.routes.api_keys.store_byor_key_in_db')
     @patch('server.routes.api_keys.generate_byor_key')
     @patch('server.routes.api_keys.delete_byor_key_from_litellm')
@@ -239,12 +247,14 @@ class TestGetLlmApiKeyForByor:
         mock_delete_key,
         mock_generate_key,
         mock_store_key,
+        mock_check_enabled,
     ):
         """Test that when an invalid key exists in database, it is regenerated."""
         # Arrange
         user_id = 'user-123'
         invalid_key = 'sk-invalid-key'
         new_key = 'sk-new-generated-key'
+        mock_check_enabled.return_value = True
         mock_get_key.return_value = invalid_key
         mock_verify_key.return_value = False
         mock_delete_key.return_value = True
@@ -256,6 +266,7 @@ class TestGetLlmApiKeyForByor:
 
         # Assert
         assert result == {'key': new_key}
+        mock_check_enabled.assert_called_once_with(user_id)
         mock_get_key.assert_called_once_with(user_id)
         mock_verify_key.assert_called_once_with(invalid_key, user_id)
         mock_delete_key.assert_called_once_with(user_id, invalid_key)
@@ -263,6 +274,7 @@ class TestGetLlmApiKeyForByor:
         mock_store_key.assert_called_once_with(user_id, new_key)
 
     @pytest.mark.asyncio
+    @patch('storage.org_service.OrgService.check_byor_export_enabled')
     @patch('server.routes.api_keys.store_byor_key_in_db')
     @patch('server.routes.api_keys.generate_byor_key')
     @patch('server.routes.api_keys.delete_byor_key_from_litellm')
@@ -275,12 +287,14 @@ class TestGetLlmApiKeyForByor:
         mock_delete_key,
         mock_generate_key,
         mock_store_key,
+        mock_check_enabled,
     ):
         """Test that even if deletion fails, regeneration still proceeds."""
         # Arrange
         user_id = 'user-123'
         invalid_key = 'sk-invalid-key'
         new_key = 'sk-new-generated-key'
+        mock_check_enabled.return_value = True
         mock_get_key.return_value = invalid_key
         mock_verify_key.return_value = False
         mock_delete_key.return_value = False  # Deletion fails
@@ -292,19 +306,22 @@ class TestGetLlmApiKeyForByor:
 
         # Assert
         assert result == {'key': new_key}
+        mock_check_enabled.assert_called_once_with(user_id)
         mock_delete_key.assert_called_once_with(user_id, invalid_key)
         mock_generate_key.assert_called_once_with(user_id)
         mock_store_key.assert_called_once_with(user_id, new_key)
 
     @pytest.mark.asyncio
+    @patch('storage.org_service.OrgService.check_byor_export_enabled')
     @patch('server.routes.api_keys.generate_byor_key')
     @patch('server.routes.api_keys.get_byor_key_from_db')
     async def test_key_generation_failure_raises_exception(
-        self, mock_get_key, mock_generate_key
+        self, mock_get_key, mock_generate_key, mock_check_enabled
     ):
         """Test that when key generation fails, an HTTPException is raised."""
         # Arrange
         user_id = 'user-123'
+        mock_check_enabled.return_value = True
         mock_get_key.return_value = None
         mock_generate_key.return_value = None
 
@@ -316,11 +333,15 @@ class TestGetLlmApiKeyForByor:
         assert 'Failed to generate new BYOR LLM API key' in exc_info.value.detail
 
     @pytest.mark.asyncio
+    @patch('storage.org_service.OrgService.check_byor_export_enabled')
     @patch('server.routes.api_keys.get_byor_key_from_db')
-    async def test_database_error_raises_exception(self, mock_get_key):
+    async def test_database_error_raises_exception(
+        self, mock_get_key, mock_check_enabled
+    ):
         """Test that database errors are properly handled."""
         # Arrange
         user_id = 'user-123'
+        mock_check_enabled.return_value = True
         mock_get_key.side_effect = Exception('Database connection error')
 
         # Act & Assert
@@ -330,6 +351,21 @@ class TestGetLlmApiKeyForByor:
         assert exc_info.value.status_code == 500
         assert 'Failed to retrieve BYOR LLM API key' in exc_info.value.detail
 
+    @pytest.mark.asyncio
+    @patch('storage.org_service.OrgService.check_byor_export_enabled')
+    async def test_byor_export_disabled_returns_402(self, mock_check_enabled):
+        """Test that when BYOR export is disabled, 402 is returned."""
+        # Arrange
+        user_id = 'user-123'
+        mock_check_enabled.return_value = False
+
+        # Act & Assert
+        with pytest.raises(HTTPException) as exc_info:
+            await get_llm_api_key_for_byor(user_id=user_id)
+
+        assert exc_info.value.status_code == 402
+        assert 'BYOR key export is not enabled' in exc_info.value.detail
+
 
 class TestDeleteByorKeyFromLitellm:
     """Test the delete_byor_key_from_litellm function with alias cleanup."""
@@ -425,3 +461,52 @@ class TestDeleteByorKeyFromLitellm:
 
         # Assert
         assert result is False
+
+
+class TestCheckByorPermitted:
+    """Test the check_byor_permitted endpoint."""
+
+    @pytest.mark.asyncio
+    @patch('storage.org_service.OrgService.check_byor_export_enabled')
+    async def test_permitted_when_enabled(self, mock_check_enabled):
+        """Test that permitted=True is returned when BYOR export is enabled."""
+        # Arrange
+        user_id = 'user-123'
+        mock_check_enabled.return_value = True
+
+        # Act
+        result = await check_byor_permitted(user_id=user_id)
+
+        # Assert
+        assert result == {'permitted': True}
+        mock_check_enabled.assert_called_once_with(user_id)
+
+    @pytest.mark.asyncio
+    @patch('storage.org_service.OrgService.check_byor_export_enabled')
+    async def test_not_permitted_when_disabled(self, mock_check_enabled):
+        """Test that permitted=False is returned when BYOR export is disabled."""
+        # Arrange
+        user_id = 'user-123'
+        mock_check_enabled.return_value = False
+
+        # Act
+        result = await check_byor_permitted(user_id=user_id)
+
+        # Assert
+        assert result == {'permitted': False}
+        mock_check_enabled.assert_called_once_with(user_id)
+
+    @pytest.mark.asyncio
+    @patch('storage.org_service.OrgService.check_byor_export_enabled')
+    async def test_error_raises_500(self, mock_check_enabled):
+        """Test that an exception raises 500 error."""
+        # Arrange
+        user_id = 'user-123'
+        mock_check_enabled.side_effect = Exception('Database error')
+
+        # Act & Assert
+        with pytest.raises(HTTPException) as exc_info:
+            await check_byor_permitted(user_id=user_id)
+
+        assert exc_info.value.status_code == 500
+        assert 'Failed to check BYOR export permission' in exc_info.value.detail

enterprise/tests/unit/server/routes/test_github_proxy.py

@@ -0,0 +1,96 @@
+from unittest.mock import patch
+from urllib.parse import parse_qs, urlparse
+
+import pytest
+from fastapi import FastAPI
+from fastapi.testclient import TestClient
+from pydantic import SecretStr
+from server.routes.github_proxy import add_github_proxy_routes
+
+
+@pytest.fixture
+def app_with_github_proxy(monkeypatch):
+    """Create a FastAPI app with github proxy routes enabled."""
+    # Enable the github proxy endpoints
+    monkeypatch.setenv('GITHUB_PROXY_ENDPOINTS', '1')
+
+    # Mock the config to have a jwt_secret
+    mock_config = type(
+        'MockConfig', (), {'jwt_secret': SecretStr('test-secret-key-for-testing')}
+    )()
+
+    app = FastAPI()
+
+    with patch('server.routes.github_proxy.GITHUB_PROXY_ENDPOINTS', True):
+        with patch('server.routes.github_proxy.config', mock_config):
+            add_github_proxy_routes(app)
+
+    # Return app and mock_config so we can use the same config in tests
+    return app, mock_config
+
+
+def test_state_compress_encrypt_and_decrypt_decompress_roundtrip(
+    app_with_github_proxy, monkeypatch
+):
+    """
+    Verify the code path used by github_proxy_start -> github_proxy_callback:
+    - compress payload, encrypt, base64-encode (what the start code does)
+    - base64-decode, decrypt, decompress (what the callback code does)
+
+    This test exercises the actual endpoints to verify the roundtrip works correctly.
+    """
+    app, mock_config = app_with_github_proxy
+    client = TestClient(app)
+
+    original_state = 'some-state-value'
+    original_redirect_uri = 'https://example.com/redirect'
+
+    # Call github_proxy_start endpoint - it should redirect to GitHub with encrypted state
+    with patch('server.routes.github_proxy.config', mock_config):
+        response = client.get(
+            '/github-proxy/test-subdomain/login/oauth/authorize',
+            params={
+                'state': original_state,
+                'redirect_uri': original_redirect_uri,
+                'client_id': 'test-client-id',
+            },
+            follow_redirects=False,
+        )
+
+    assert response.status_code == 307
+    redirect_url = response.headers['location']
+
+    # Verify it redirects to GitHub
+    assert redirect_url.startswith('https://github.com/login/oauth/authorize')
+
+    # Parse the redirect URL to get the encrypted state
+    parsed = urlparse(redirect_url)
+    query_params = parse_qs(parsed.query)
+    encrypted_state = query_params['state'][0]
+
+    # The redirect_uri should now point to our callback
+    assert 'github-proxy/callback' in query_params['redirect_uri'][0]
+
+    # Now simulate GitHub calling back with this encrypted state
+    with patch('server.routes.github_proxy.config', mock_config):
+        callback_response = client.get(
+            '/github-proxy/callback',
+            params={
+                'state': encrypted_state,
+                'code': 'test-auth-code',
+            },
+            follow_redirects=False,
+        )
+
+    assert callback_response.status_code == 307
+    final_redirect = callback_response.headers['location']
+
+    # Verify the callback redirects to the original redirect_uri
+    assert final_redirect.startswith(original_redirect_uri)
+
+    # Parse the final redirect to verify the state was decrypted correctly
+    final_parsed = urlparse(final_redirect)
+    final_params = parse_qs(final_parsed.query)
+
+    assert final_params['state'][0] == original_state
+    assert final_params['code'][0] == 'test-auth-code'

enterprise/tests/unit/server/routes/test_oauth_device.py

@@ -1,7 +1,7 @@
 """Unit tests for OAuth2 Device Flow endpoints."""
 
 from datetime import UTC, datetime, timedelta
-from unittest.mock import MagicMock, patch
+from unittest.mock import AsyncMock, MagicMock, patch
 
 import pytest
 from fastapi import HTTPException, Request
@@ -22,8 +22,10 @@ def mock_device_code_store():
 
 @pytest.fixture
 def mock_api_key_store():
-    """Mock API key store."""
-    return MagicMock()
+    """Mock API key store with async create_api_key."""
+    mock = MagicMock()
+    mock.create_api_key = AsyncMock()
+    return mock
 
 
 @pytest.fixture
@@ -204,8 +206,9 @@ class TestDeviceVerificationAuthenticated:
         mock_store.get_by_user_code.return_value = mock_device
         mock_store.authorize_device_code.return_value = True
 
-        # Mock API key store
+        # Mock API key store with async create_api_key
         mock_api_key_store = MagicMock()
+        mock_api_key_store.create_api_key = AsyncMock()
         mock_api_key_class.get_instance.return_value = mock_api_key_store
 
         result = await device_verification_authenticated(
@@ -228,8 +231,9 @@ class TestDeviceVerificationAuthenticated:
     @patch('server.routes.oauth_device.device_code_store')
     async def test_multiple_device_authentication(self, mock_store, mock_api_key_class):
         """Test that multiple devices can authenticate simultaneously."""
-        # Mock API key store
+        # Mock API key store with async create_api_key
         mock_api_key_store = MagicMock()
+        mock_api_key_store.create_api_key = AsyncMock()
         mock_api_key_class.get_instance.return_value = mock_api_key_store
 
         # Simulate two different devices
@@ -486,8 +490,9 @@ class TestDeviceVerificationTransactionIntegrity:
         mock_store.get_by_user_code.return_value = mock_device
         mock_store.authorize_device_code.return_value = False  # Authorization fails
 
-        # Mock API key store
+        # Mock API key store with async create_api_key
         mock_api_key_store = MagicMock()
+        mock_api_key_store.create_api_key = AsyncMock()
         mock_api_key_class.get_instance.return_value = mock_api_key_store
 
         # Should raise HTTPException due to authorization failure
@@ -518,9 +523,11 @@ class TestDeviceVerificationTransactionIntegrity:
         mock_store.authorize_device_code.return_value = True  # Authorization succeeds
         mock_store.deny_device_code.return_value = True  # Cleanup succeeds
 
-        # Mock API key store to fail on creation
+        # Mock API key store to fail on creation (async)
         mock_api_key_store = MagicMock()
-        mock_api_key_store.create_api_key.side_effect = Exception('Database error')
+        mock_api_key_store.create_api_key = AsyncMock(
+            side_effect=Exception('Database error')
+        )
         mock_api_key_class.get_instance.return_value = mock_api_key_store
 
         # Should raise HTTPException due to API key creation failure
@@ -558,9 +565,11 @@ class TestDeviceVerificationTransactionIntegrity:
             'Cleanup failed'
         )  # Cleanup fails
 
-        # Mock API key store to fail on creation
+        # Mock API key store to fail on creation (async)
         mock_api_key_store = MagicMock()
-        mock_api_key_store.create_api_key.side_effect = Exception('Database error')
+        mock_api_key_store.create_api_key = AsyncMock(
+            side_effect=Exception('Database error')
+        )
         mock_api_key_class.get_instance.return_value = mock_api_key_store
 
         # Should still raise HTTPException for the original API key creation failure
@@ -589,8 +598,9 @@ class TestDeviceVerificationTransactionIntegrity:
         mock_store.get_by_user_code.return_value = mock_device
         mock_store.authorize_device_code.return_value = True  # Authorization succeeds
 
-        # Mock API key store
+        # Mock API key store with async create_api_key
         mock_api_key_store = MagicMock()
+        mock_api_key_store.create_api_key = AsyncMock()
         mock_api_key_class.get_instance.return_value = mock_api_key_store
 
         result = await device_verification_authenticated(

enterprise/tests/unit/sync/test_resend_keycloak.py

@@ -0,0 +1,117 @@
+"""Tests for resend_keycloak email validation."""
+
+from sync.resend_keycloak import is_valid_email
+
+
+class TestIsValidEmail:
+    """Test cases for is_valid_email function."""
+
+    def test_valid_simple_email(self):
+        """Test that a simple valid email passes validation."""
+        assert is_valid_email('user@example.com') is True
+
+    def test_valid_email_with_plus(self):
+        """Test that email with + modifier passes validation."""
+        assert is_valid_email('user+tag@example.com') is True
+
+    def test_valid_email_with_dots(self):
+        """Test that email with dots in local part passes validation."""
+        assert is_valid_email('first.last@example.com') is True
+
+    def test_valid_email_with_numbers(self):
+        """Test that email with numbers passes validation."""
+        assert is_valid_email('user123@example.com') is True
+
+    def test_valid_email_with_subdomain(self):
+        """Test that email with subdomain passes validation."""
+        assert is_valid_email('user@mail.example.com') is True
+
+    def test_valid_email_with_hyphen_domain(self):
+        """Test that email with hyphen in domain passes validation."""
+        assert is_valid_email('user@example-site.com') is True
+
+    def test_valid_email_with_underscore(self):
+        """Test that email with underscore passes validation."""
+        assert is_valid_email('user_name@example.com') is True
+
+    def test_valid_email_with_percent(self):
+        """Test that email with percent sign passes validation."""
+        assert is_valid_email('user%name@example.com') is True
+
+    def test_invalid_email_with_exclamation(self):
+        """Test that email with exclamation mark fails validation.
+
+        This is the specific case from the bug report:
+        ethanjames3713+!@gmail.com
+        """
+        assert is_valid_email('ethanjames3713+!@gmail.com') is False
+
+    def test_invalid_email_with_special_chars(self):
+        """Test that email with other special characters fails validation."""
+        assert is_valid_email('user!name@example.com') is False
+        assert is_valid_email('user#name@example.com') is False
+        assert is_valid_email('user$name@example.com') is False
+        assert is_valid_email('user&name@example.com') is False
+        assert is_valid_email("user'name@example.com") is False
+        assert is_valid_email('user*name@example.com') is False
+        assert is_valid_email('user=name@example.com') is False
+        assert is_valid_email('user^name@example.com') is False
+        assert is_valid_email('user`name@example.com') is False
+        assert is_valid_email('user{name@example.com') is False
+        assert is_valid_email('user|name@example.com') is False
+        assert is_valid_email('user}name@example.com') is False
+        assert is_valid_email('user~name@example.com') is False
+
+    def test_invalid_email_no_at_symbol(self):
+        """Test that email without @ symbol fails validation."""
+        assert is_valid_email('userexample.com') is False
+
+    def test_invalid_email_no_domain(self):
+        """Test that email without domain fails validation."""
+        assert is_valid_email('user@') is False
+
+    def test_invalid_email_no_local_part(self):
+        """Test that email without local part fails validation."""
+        assert is_valid_email('@example.com') is False
+
+    def test_invalid_email_no_tld(self):
+        """Test that email without TLD fails validation."""
+        assert is_valid_email('user@example') is False
+
+    def test_invalid_email_single_char_tld(self):
+        """Test that email with single character TLD fails validation."""
+        assert is_valid_email('user@example.c') is False
+
+    def test_invalid_email_empty_string(self):
+        """Test that empty string fails validation."""
+        assert is_valid_email('') is False
+
+    def test_invalid_email_none(self):
+        """Test that None fails validation."""
+        assert is_valid_email(None) is False
+
+    def test_invalid_email_whitespace(self):
+        """Test that email with whitespace fails validation."""
+        assert is_valid_email('user @example.com') is False
+        assert is_valid_email('user@ example.com') is False
+        assert is_valid_email(' user@example.com') is False
+        assert is_valid_email('user@example.com ') is False
+
+    def test_invalid_email_double_at(self):
+        """Test that email with double @ fails validation."""
+        assert is_valid_email('user@@example.com') is False
+
+    def test_email_double_dot_domain(self):
+        """Test email with double dot in domain.
+
+        Note: The regex allows this as it's technically valid in some edge cases,
+        and Resend's API may accept it. The main goal is to reject special
+        characters like ! that Resend definitely rejects.
+        """
+        # This is allowed by our regex - Resend may or may not accept it
+        assert is_valid_email('user@example..com') is True
+
+    def test_case_insensitive_validation(self):
+        """Test that validation works for uppercase emails."""
+        assert is_valid_email('USER@EXAMPLE.COM') is True
+        assert is_valid_email('User@Example.Com') is True

enterprise/tests/unit/test_api_key_store.py

@@ -32,6 +32,11 @@ def api_key_store(mock_session_maker):
     return ApiKeyStore(mock_session_maker)
 
 
+def run_sync(func, *args, **kwargs):
+    """Helper to execute sync functions directly (mocks call_sync_from_async)."""
+    return func(*args, **kwargs)
+
+
 def test_generate_api_key(api_key_store):
     """Test that generate_api_key returns a string with sk-oh- prefix and expected length."""
     key = api_key_store.generate_api_key(length=32)
@@ -41,8 +46,12 @@ def test_generate_api_key(api_key_store):
     assert len(key) == len('sk-oh-') + 32
 
 
-@patch('storage.api_key_store.UserStore.get_user_by_id')
-def test_create_api_key(mock_get_user, api_key_store, mock_session, mock_user):
+@pytest.mark.asyncio
+@patch('storage.api_key_store.call_sync_from_async', side_effect=run_sync)
+@patch('storage.api_key_store.UserStore.get_user_by_id_async')
+async def test_create_api_key(
+    mock_get_user, mock_call_sync, api_key_store, mock_session, mock_user
+):
     """Test creating an API key."""
     # Setup
     user_id = 'test-user-123'
@@ -51,7 +60,7 @@ def test_create_api_key(mock_get_user, api_key_store, mock_session, mock_user):
     api_key_store.generate_api_key = MagicMock(return_value='test-api-key')
 
     # Execute
-    result = api_key_store.create_api_key(user_id, name)
+    result = await api_key_store.create_api_key(user_id, name)
 
     # Verify
     assert result == 'test-api-key'
@@ -219,8 +228,12 @@ def test_delete_api_key_by_id(api_key_store, mock_session):
     mock_session.commit.assert_called_once()
 
 
-@patch('storage.api_key_store.UserStore.get_user_by_id')
-def test_list_api_keys(mock_get_user, api_key_store, mock_session, mock_user):
+@pytest.mark.asyncio
+@patch('storage.api_key_store.call_sync_from_async', side_effect=run_sync)
+@patch('storage.api_key_store.UserStore.get_user_by_id_async')
+async def test_list_api_keys(
+    mock_get_user, mock_call_sync, api_key_store, mock_session, mock_user
+):
     """Test listing API keys for a user."""
     # Setup
     user_id = 'test-user-123'
@@ -247,7 +260,7 @@ def test_list_api_keys(mock_get_user, api_key_store, mock_session, mock_user):
     mock_filter_org.all.return_value = [mock_key1, mock_key2]
 
     # Execute
-    result = api_key_store.list_api_keys(user_id)
+    result = await api_key_store.list_api_keys(user_id)
 
     # Verify
     mock_get_user.assert_called_once_with(user_id)
@@ -265,8 +278,12 @@ def test_list_api_keys(mock_get_user, api_key_store, mock_session, mock_user):
     assert result[1]['expires_at'] is None
 
 
-@patch('storage.api_key_store.UserStore.get_user_by_id')
-def test_retrieve_mcp_api_key(mock_get_user, api_key_store, mock_session, mock_user):
+@pytest.mark.asyncio
+@patch('storage.api_key_store.call_sync_from_async', side_effect=run_sync)
+@patch('storage.api_key_store.UserStore.get_user_by_id_async')
+async def test_retrieve_mcp_api_key(
+    mock_get_user, mock_call_sync, api_key_store, mock_session, mock_user
+):
     """Test retrieving MCP API key for a user."""
     # Setup
     user_id = 'test-user-123'
@@ -287,16 +304,18 @@ def test_retrieve_mcp_api_key(mock_get_user, api_key_store, mock_session, mock_u
     mock_filter_org.all.return_value = [mock_other_key, mock_mcp_key]
 
     # Execute
-    result = api_key_store.retrieve_mcp_api_key(user_id)
+    result = await api_key_store.retrieve_mcp_api_key(user_id)
 
     # Verify
     mock_get_user.assert_called_once_with(user_id)
     assert result == 'mcp-test-key'
 
 
-@patch('storage.api_key_store.UserStore.get_user_by_id')
-def test_retrieve_mcp_api_key_not_found(
-    mock_get_user, api_key_store, mock_session, mock_user
+@pytest.mark.asyncio
+@patch('storage.api_key_store.call_sync_from_async', side_effect=run_sync)
+@patch('storage.api_key_store.UserStore.get_user_by_id_async')
+async def test_retrieve_mcp_api_key_not_found(
+    mock_get_user, mock_call_sync, api_key_store, mock_session, mock_user
 ):
     """Test retrieving MCP API key when none exists."""
     # Setup
@@ -314,7 +333,7 @@ def test_retrieve_mcp_api_key_not_found(
     mock_filter_org.all.return_value = [mock_other_key]
 
     # Execute
-    result = api_key_store.retrieve_mcp_api_key(user_id)
+    result = await api_key_store.retrieve_mcp_api_key(user_id)
 
     # Verify
     mock_get_user.assert_called_once_with(user_id)

enterprise/tests/unit/test_auth_routes.py

@@ -153,6 +153,7 @@ async def test_keycloak_callback_user_not_allowed(mock_request):
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
         mock_user_store.migrate_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         mock_verifier.is_active.return_value = True
         mock_verifier.is_user_allowed.return_value = False
@@ -188,6 +189,7 @@ async def test_keycloak_callback_success_with_valid_offline_token(mock_request):
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
         mock_user_store.migrate_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         mock_token_manager.get_keycloak_tokens = AsyncMock(
             return_value=('test_access_token', 'test_refresh_token')
@@ -259,6 +261,7 @@ async def test_keycloak_callback_email_not_verified(mock_request):
         mock_user.current_org_id = 'test_org_id'
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         # Act
         result = await keycloak_callback(
@@ -306,6 +309,7 @@ async def test_keycloak_callback_email_not_verified_missing_field(mock_request):
         mock_user.current_org_id = 'test_org_id'
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         # Act
         result = await keycloak_callback(
@@ -347,6 +351,7 @@ async def test_keycloak_callback_success_without_offline_token(mock_request):
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
         mock_user_store.migrate_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         mock_token_manager.get_keycloak_tokens = AsyncMock(
             return_value=('test_access_token', 'test_refresh_token')
@@ -581,6 +586,7 @@ async def test_keycloak_callback_blocked_email_domain(mock_request):
         mock_user.current_org_id = 'test_org_id'
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         mock_domain_blocker.is_active.return_value = True
         mock_domain_blocker.is_domain_blocked.return_value = True
@@ -644,6 +650,7 @@ async def test_keycloak_callback_allowed_email_domain(mock_request):
         mock_user.accepted_tos = '2025-01-01'
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         mock_domain_blocker.is_active.return_value = True
         mock_domain_blocker.is_domain_blocked.return_value = False
@@ -707,6 +714,7 @@ async def test_keycloak_callback_domain_blocking_inactive(mock_request):
         mock_user.accepted_tos = '2025-01-01'
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         mock_domain_blocker.is_active.return_value = False
         mock_domain_blocker.is_domain_blocked.return_value = False
@@ -768,6 +776,7 @@ async def test_keycloak_callback_missing_email(mock_request):
         mock_user.accepted_tos = '2025-01-01'
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         mock_domain_blocker.is_active.return_value = True
 
@@ -813,6 +822,7 @@ async def test_keycloak_callback_duplicate_email_detected(mock_request):
         mock_user.current_org_id = 'test_org_id'
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         # Act
         result = await keycloak_callback(
@@ -857,6 +867,7 @@ async def test_keycloak_callback_duplicate_email_deletion_fails(mock_request):
         mock_user.current_org_id = 'test_org_id'
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         # Act
         result = await keycloak_callback(
@@ -914,6 +925,7 @@ async def test_keycloak_callback_duplicate_check_exception(mock_request):
         mock_user.accepted_tos = '2025-01-01'
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         mock_verifier.is_active.return_value = True
         mock_verifier.is_user_allowed.return_value = True
@@ -971,6 +983,7 @@ async def test_keycloak_callback_no_duplicate_email(mock_request):
         mock_user.accepted_tos = '2025-01-01'
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         mock_verifier.is_active.return_value = True
         mock_verifier.is_user_allowed.return_value = True
@@ -1031,6 +1044,7 @@ async def test_keycloak_callback_no_email_in_user_info(mock_request):
         mock_user.accepted_tos = '2025-01-01'
         mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
         mock_user_store.create_user = AsyncMock(return_value=mock_user)
+        mock_user_store.backfill_contact_name = AsyncMock()
 
         mock_verifier.is_active.return_value = True
         mock_verifier.is_user_allowed.return_value = True
@@ -1187,6 +1201,7 @@ class TestKeycloakCallbackRecaptcha:
             mock_user.accepted_tos = '2025-01-01'
             mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
             mock_user_store.create_user = AsyncMock(return_value=mock_user)
+            mock_user_store.backfill_contact_name = AsyncMock()
 
             mock_verifier.is_active.return_value = True
             mock_verifier.is_user_allowed.return_value = True
@@ -1251,6 +1266,7 @@ class TestKeycloakCallbackRecaptcha:
             mock_user.current_org_id = 'test_org_id'
             mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
             mock_user_store.create_user = AsyncMock(return_value=mock_user)
+            mock_user_store.backfill_contact_name = AsyncMock()
 
             mock_domain_blocker.is_domain_blocked.return_value = False
 
@@ -1333,6 +1349,7 @@ class TestKeycloakCallbackRecaptcha:
             mock_user.accepted_tos = '2025-01-01'
             mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
             mock_user_store.create_user = AsyncMock(return_value=mock_user)
+            mock_user_store.backfill_contact_name = AsyncMock()
 
             mock_verifier.is_active.return_value = True
             mock_verifier.is_user_allowed.return_value = True
@@ -1420,6 +1437,7 @@ class TestKeycloakCallbackRecaptcha:
             mock_user.accepted_tos = '2025-01-01'
             mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
             mock_user_store.create_user = AsyncMock(return_value=mock_user)
+            mock_user_store.backfill_contact_name = AsyncMock()
 
             mock_verifier.is_active.return_value = True
             mock_verifier.is_user_allowed.return_value = True
@@ -1504,6 +1522,7 @@ class TestKeycloakCallbackRecaptcha:
             mock_user.accepted_tos = '2025-01-01'
             mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
             mock_user_store.create_user = AsyncMock(return_value=mock_user)
+            mock_user_store.backfill_contact_name = AsyncMock()
 
             mock_verifier.is_active.return_value = True
             mock_verifier.is_user_allowed.return_value = True
@@ -1587,6 +1606,7 @@ class TestKeycloakCallbackRecaptcha:
             mock_user.accepted_tos = '2025-01-01'
             mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
             mock_user_store.create_user = AsyncMock(return_value=mock_user)
+            mock_user_store.backfill_contact_name = AsyncMock()
 
             mock_verifier.is_active.return_value = True
             mock_verifier.is_user_allowed.return_value = True
@@ -1667,6 +1687,7 @@ class TestKeycloakCallbackRecaptcha:
             mock_user.accepted_tos = '2025-01-01'
             mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
             mock_user_store.create_user = AsyncMock(return_value=mock_user)
+            mock_user_store.backfill_contact_name = AsyncMock()
 
             mock_verifier.is_active.return_value = True
             mock_verifier.is_user_allowed.return_value = True
@@ -1733,6 +1754,7 @@ class TestKeycloakCallbackRecaptcha:
             mock_user.accepted_tos = '2025-01-01'
             mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
             mock_user_store.create_user = AsyncMock(return_value=mock_user)
+            mock_user_store.backfill_contact_name = AsyncMock()
 
             mock_verifier.is_active.return_value = True
             mock_verifier.is_user_allowed.return_value = True
@@ -1805,6 +1827,7 @@ class TestKeycloakCallbackRecaptcha:
             mock_user.accepted_tos = '2025-01-01'
             mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
             mock_user_store.create_user = AsyncMock(return_value=mock_user)
+            mock_user_store.backfill_contact_name = AsyncMock()
 
             mock_verifier.is_active.return_value = True
             mock_verifier.is_user_allowed.return_value = True
@@ -1875,6 +1898,7 @@ class TestKeycloakCallbackRecaptcha:
             mock_user.current_org_id = 'test_org_id'
             mock_user_store.get_user_by_id_async = AsyncMock(return_value=mock_user)
             mock_user_store.create_user = AsyncMock(return_value=mock_user)
+            mock_user_store.backfill_contact_name = AsyncMock()
 
             mock_domain_blocker.is_domain_blocked.return_value = False
 

enterprise/tests/unit/test_authorization.py

@@ -0,0 +1,491 @@
+"""
+Unit tests for role-based authorization (authorization.py).
+
+Tests the FastAPI dependencies that validate user roles within organizations.
+"""
+
+from unittest.mock import MagicMock, patch
+from uuid import uuid4
+
+import pytest
+from fastapi import HTTPException
+from server.auth.authorization import (
+    ROLE_HIERARCHY,
+    OrgRole,
+    get_user_org_role,
+    has_required_role,
+    require_org_admin,
+    require_org_owner,
+    require_org_role,
+    require_org_user,
+)
+
+# =============================================================================
+# Tests for OrgRole enum
+# =============================================================================
+
+
+class TestOrgRole:
+    """Tests for OrgRole enum."""
+
+    def test_org_role_values(self):
+        """
+        GIVEN: OrgRole enum
+        WHEN: Accessing role values
+        THEN: All expected roles exist with correct string values
+        """
+        assert OrgRole.OWNER.value == 'owner'
+        assert OrgRole.ADMIN.value == 'admin'
+        assert OrgRole.USER.value == 'user'
+
+    def test_org_role_from_string(self):
+        """
+        GIVEN: Valid role string
+        WHEN: Creating OrgRole from string
+        THEN: Correct enum value is returned
+        """
+        assert OrgRole('owner') == OrgRole.OWNER
+        assert OrgRole('admin') == OrgRole.ADMIN
+        assert OrgRole('user') == OrgRole.USER
+
+    def test_org_role_invalid_string(self):
+        """
+        GIVEN: Invalid role string
+        WHEN: Creating OrgRole from string
+        THEN: ValueError is raised
+        """
+        with pytest.raises(ValueError):
+            OrgRole('invalid_role')
+
+
+# =============================================================================
+# Tests for role hierarchy
+# =============================================================================
+
+
+class TestRoleHierarchy:
+    """Tests for role hierarchy constants."""
+
+    def test_owner_highest_rank(self):
+        """
+        GIVEN: Role hierarchy
+        WHEN: Comparing role ranks
+        THEN: Owner has highest rank
+        """
+        assert ROLE_HIERARCHY[OrgRole.OWNER] > ROLE_HIERARCHY[OrgRole.ADMIN]
+        assert ROLE_HIERARCHY[OrgRole.OWNER] > ROLE_HIERARCHY[OrgRole.USER]
+
+    def test_admin_middle_rank(self):
+        """
+        GIVEN: Role hierarchy
+        WHEN: Comparing role ranks
+        THEN: Admin is between owner and user
+        """
+        assert ROLE_HIERARCHY[OrgRole.ADMIN] > ROLE_HIERARCHY[OrgRole.USER]
+        assert ROLE_HIERARCHY[OrgRole.ADMIN] < ROLE_HIERARCHY[OrgRole.OWNER]
+
+    def test_user_lowest_rank(self):
+        """
+        GIVEN: Role hierarchy
+        WHEN: Comparing role ranks
+        THEN: User has lowest rank
+        """
+        assert ROLE_HIERARCHY[OrgRole.USER] < ROLE_HIERARCHY[OrgRole.ADMIN]
+        assert ROLE_HIERARCHY[OrgRole.USER] < ROLE_HIERARCHY[OrgRole.OWNER]
+
+
+# =============================================================================
+# Tests for has_required_role function
+# =============================================================================
+
+
+class TestHasRequiredRole:
+    """Tests for has_required_role function."""
+
+    def test_owner_has_owner_role(self):
+        """
+        GIVEN: User with owner role
+        WHEN: Checking for owner requirement
+        THEN: Returns True
+        """
+        assert has_required_role('owner', OrgRole.OWNER) is True
+
+    def test_owner_has_admin_role(self):
+        """
+        GIVEN: User with owner role
+        WHEN: Checking for admin requirement
+        THEN: Returns True (owner > admin)
+        """
+        assert has_required_role('owner', OrgRole.ADMIN) is True
+
+    def test_owner_has_user_role(self):
+        """
+        GIVEN: User with owner role
+        WHEN: Checking for user requirement
+        THEN: Returns True (owner > user)
+        """
+        assert has_required_role('owner', OrgRole.USER) is True
+
+    def test_admin_has_admin_role(self):
+        """
+        GIVEN: User with admin role
+        WHEN: Checking for admin requirement
+        THEN: Returns True
+        """
+        assert has_required_role('admin', OrgRole.ADMIN) is True
+
+    def test_admin_has_user_role(self):
+        """
+        GIVEN: User with admin role
+        WHEN: Checking for user requirement
+        THEN: Returns True (admin > user)
+        """
+        assert has_required_role('admin', OrgRole.USER) is True
+
+    def test_admin_lacks_owner_role(self):
+        """
+        GIVEN: User with admin role
+        WHEN: Checking for owner requirement
+        THEN: Returns False (admin < owner)
+        """
+        assert has_required_role('admin', OrgRole.OWNER) is False
+
+    def test_user_has_user_role(self):
+        """
+        GIVEN: User with user role
+        WHEN: Checking for user requirement
+        THEN: Returns True
+        """
+        assert has_required_role('user', OrgRole.USER) is True
+
+    def test_user_lacks_admin_role(self):
+        """
+        GIVEN: User with user role
+        WHEN: Checking for admin requirement
+        THEN: Returns False (user < admin)
+        """
+        assert has_required_role('user', OrgRole.ADMIN) is False
+
+    def test_user_lacks_owner_role(self):
+        """
+        GIVEN: User with user role
+        WHEN: Checking for owner requirement
+        THEN: Returns False (user < owner)
+        """
+        assert has_required_role('user', OrgRole.OWNER) is False
+
+    def test_invalid_role_returns_false(self):
+        """
+        GIVEN: Invalid role string
+        WHEN: Checking for any requirement
+        THEN: Returns False
+        """
+        assert has_required_role('invalid_role', OrgRole.USER) is False
+        assert has_required_role('invalid_role', OrgRole.ADMIN) is False
+        assert has_required_role('invalid_role', OrgRole.OWNER) is False
+
+
+# =============================================================================
+# Tests for get_user_org_role function
+# =============================================================================
+
+
+class TestGetUserOrgRole:
+    """Tests for get_user_org_role function."""
+
+    def test_returns_role_when_member_exists(self):
+        """
+        GIVEN: User is a member of organization with role
+        WHEN: get_user_org_role is called
+        THEN: Role name is returned
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        mock_org_member = MagicMock()
+        mock_org_member.role_id = 1
+
+        mock_role = MagicMock()
+        mock_role.name = 'admin'
+
+        with (
+            patch(
+                'server.auth.authorization.OrgMemberStore.get_org_member',
+                return_value=mock_org_member,
+            ),
+            patch(
+                'server.auth.authorization.RoleStore.get_role_by_id',
+                return_value=mock_role,
+            ),
+        ):
+            result = get_user_org_role(user_id, org_id)
+            assert result == 'admin'
+
+    def test_returns_none_when_not_member(self):
+        """
+        GIVEN: User is not a member of organization
+        WHEN: get_user_org_role is called
+        THEN: None is returned
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        with patch(
+            'server.auth.authorization.OrgMemberStore.get_org_member',
+            return_value=None,
+        ):
+            result = get_user_org_role(user_id, org_id)
+            assert result is None
+
+    def test_returns_none_when_role_not_found(self):
+        """
+        GIVEN: User is member but role not found
+        WHEN: get_user_org_role is called
+        THEN: None is returned
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        mock_org_member = MagicMock()
+        mock_org_member.role_id = 999  # Non-existent role
+
+        with (
+            patch(
+                'server.auth.authorization.OrgMemberStore.get_org_member',
+                return_value=mock_org_member,
+            ),
+            patch(
+                'server.auth.authorization.RoleStore.get_role_by_id',
+                return_value=None,
+            ),
+        ):
+            result = get_user_org_role(user_id, org_id)
+            assert result is None
+
+
+# =============================================================================
+# Tests for require_org_role dependency
+# =============================================================================
+
+
+class TestRequireOrgRole:
+    """Tests for require_org_role dependency factory."""
+
+    @pytest.mark.asyncio
+    async def test_returns_user_id_when_authorized(self):
+        """
+        GIVEN: User with sufficient role
+        WHEN: Role checker is called
+        THEN: User ID is returned
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        with patch(
+            'server.auth.authorization.get_user_org_role',
+            return_value='admin',
+        ):
+            role_checker = require_org_role(OrgRole.USER)
+            result = await role_checker(org_id=org_id, user_id=user_id)
+            assert result == user_id
+
+    @pytest.mark.asyncio
+    async def test_raises_401_when_not_authenticated(self):
+        """
+        GIVEN: No user ID (not authenticated)
+        WHEN: Role checker is called
+        THEN: 401 Unauthorized is raised
+        """
+        org_id = uuid4()
+
+        role_checker = require_org_role(OrgRole.USER)
+        with pytest.raises(HTTPException) as exc_info:
+            await role_checker(org_id=org_id, user_id=None)
+
+        assert exc_info.value.status_code == 401
+        assert 'not authenticated' in exc_info.value.detail.lower()
+
+    @pytest.mark.asyncio
+    async def test_raises_403_when_not_member(self):
+        """
+        GIVEN: User is not a member of organization
+        WHEN: Role checker is called
+        THEN: 403 Forbidden is raised
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        with patch(
+            'server.auth.authorization.get_user_org_role',
+            return_value=None,
+        ):
+            role_checker = require_org_role(OrgRole.USER)
+            with pytest.raises(HTTPException) as exc_info:
+                await role_checker(org_id=org_id, user_id=user_id)
+
+            assert exc_info.value.status_code == 403
+            assert 'not a member' in exc_info.value.detail.lower()
+
+    @pytest.mark.asyncio
+    async def test_raises_403_when_insufficient_role(self):
+        """
+        GIVEN: User with insufficient role
+        WHEN: Role checker is called
+        THEN: 403 Forbidden is raised
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        with patch(
+            'server.auth.authorization.get_user_org_role',
+            return_value='user',
+        ):
+            role_checker = require_org_role(OrgRole.ADMIN)
+            with pytest.raises(HTTPException) as exc_info:
+                await role_checker(org_id=org_id, user_id=user_id)
+
+            assert exc_info.value.status_code == 403
+            assert 'admin' in exc_info.value.detail.lower()
+
+    @pytest.mark.asyncio
+    async def test_owner_satisfies_admin_requirement(self):
+        """
+        GIVEN: User with owner role
+        WHEN: Admin role is required
+        THEN: User ID is returned (owner > admin)
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        with patch(
+            'server.auth.authorization.get_user_org_role',
+            return_value='owner',
+        ):
+            role_checker = require_org_role(OrgRole.ADMIN)
+            result = await role_checker(org_id=org_id, user_id=user_id)
+            assert result == user_id
+
+    @pytest.mark.asyncio
+    async def test_logs_warning_on_insufficient_role(self):
+        """
+        GIVEN: User with insufficient role
+        WHEN: Role checker is called
+        THEN: Warning is logged with details
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        with (
+            patch(
+                'server.auth.authorization.get_user_org_role',
+                return_value='user',
+            ),
+            patch('server.auth.authorization.logger') as mock_logger,
+        ):
+            role_checker = require_org_role(OrgRole.OWNER)
+            with pytest.raises(HTTPException):
+                await role_checker(org_id=org_id, user_id=user_id)
+
+            mock_logger.warning.assert_called()
+            call_args = mock_logger.warning.call_args
+            assert call_args[1]['extra']['user_id'] == user_id
+            assert call_args[1]['extra']['user_role'] == 'user'
+            assert call_args[1]['extra']['required_role'] == 'owner'
+
+
+# =============================================================================
+# Tests for convenience dependencies
+# =============================================================================
+
+
+class TestConvenienceDependencies:
+    """Tests for pre-configured convenience dependencies."""
+
+    @pytest.mark.asyncio
+    async def test_require_org_user_allows_user(self):
+        """
+        GIVEN: User with user role
+        WHEN: require_org_user is used
+        THEN: User ID is returned
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        with patch(
+            'server.auth.authorization.get_user_org_role',
+            return_value='user',
+        ):
+            result = await require_org_user(org_id=org_id, user_id=user_id)
+            assert result == user_id
+
+    @pytest.mark.asyncio
+    async def test_require_org_admin_allows_admin(self):
+        """
+        GIVEN: User with admin role
+        WHEN: require_org_admin is used
+        THEN: User ID is returned
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        with patch(
+            'server.auth.authorization.get_user_org_role',
+            return_value='admin',
+        ):
+            result = await require_org_admin(org_id=org_id, user_id=user_id)
+            assert result == user_id
+
+    @pytest.mark.asyncio
+    async def test_require_org_admin_rejects_user(self):
+        """
+        GIVEN: User with user role
+        WHEN: require_org_admin is used
+        THEN: 403 Forbidden is raised
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        with patch(
+            'server.auth.authorization.get_user_org_role',
+            return_value='user',
+        ):
+            with pytest.raises(HTTPException) as exc_info:
+                await require_org_admin(org_id=org_id, user_id=user_id)
+
+            assert exc_info.value.status_code == 403
+
+    @pytest.mark.asyncio
+    async def test_require_org_owner_allows_owner(self):
+        """
+        GIVEN: User with owner role
+        WHEN: require_org_owner is used
+        THEN: User ID is returned
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        with patch(
+            'server.auth.authorization.get_user_org_role',
+            return_value='owner',
+        ):
+            result = await require_org_owner(org_id=org_id, user_id=user_id)
+            assert result == user_id
+
+    @pytest.mark.asyncio
+    async def test_require_org_owner_rejects_admin(self):
+        """
+        GIVEN: User with admin role
+        WHEN: require_org_owner is used
+        THEN: 403 Forbidden is raised
+        """
+        user_id = str(uuid4())
+        org_id = uuid4()
+
+        with patch(
+            'server.auth.authorization.get_user_org_role',
+            return_value='admin',
+        ):
+            with pytest.raises(HTTPException) as exc_info:
+                await require_org_owner(org_id=org_id, user_id=user_id)
+
+            assert exc_info.value.status_code == 403

enterprise/tests/unit/test_billing.py

@@ -163,7 +163,7 @@ async def test_create_checkout_session_stripe_error(
             'server.auth.token_manager.TokenManager.get_user_info_from_user_id',
             AsyncMock(return_value={'email': 'testy@tester.com'}),
         ),
-        patch('server.routes.billing.validate_saas_environment'),
+        patch('server.routes.billing.validate_billing_enabled'),
     ):
         await create_checkout_session(
             CreateCheckoutSessionRequest(amount=25), mock_checkout_request, 'mock_user'
@@ -204,7 +204,7 @@ async def test_create_checkout_session_success(session_maker, mock_checkout_requ
             'server.auth.token_manager.TokenManager.get_user_info_from_user_id',
             AsyncMock(return_value={'email': 'testy@tester.com'}),
         ),
-        patch('server.routes.billing.validate_saas_environment'),
+        patch('server.routes.billing.validate_billing_enabled'),
     ):
         mock_db_session = MagicMock()
         mock_session_maker.return_value.__enter__.return_value = mock_db_session
@@ -236,8 +236,8 @@ async def test_create_checkout_session_success(session_maker, mock_checkout_requ
             mode='payment',
             payment_method_types=['card'],
             saved_payment_method_options={'payment_method_save': 'enabled'},
-            success_url='http://test.com/api/billing/success?session_id={CHECKOUT_SESSION_ID}',
-            cancel_url='http://test.com/api/billing/cancel?session_id={CHECKOUT_SESSION_ID}',
+            success_url='https://test.com/api/billing/success?session_id={CHECKOUT_SESSION_ID}',
+            cancel_url='https://test.com/api/billing/cancel?session_id={CHECKOUT_SESSION_ID}',
         )
 
         # Verify database session creation
@@ -299,6 +299,8 @@ async def test_success_callback_success():
     mock_billing_session.status = 'in_progress'
     mock_billing_session.user_id = 'mock_user'
 
+    mock_org = MagicMock()
+
     with (
         patch('server.routes.billing.session_maker') as mock_session_maker,
         patch('stripe.checkout.Session.retrieve') as mock_stripe_retrieve,
@@ -319,7 +321,17 @@ async def test_success_callback_success():
         ) as mock_update_budget,
     ):
         mock_db_session = MagicMock()
+        # First query: BillingSession (query().filter().filter().first())
         mock_db_session.query.return_value.filter.return_value.filter.return_value.first.return_value = mock_billing_session
+        # Second query: Org (query().filter().first()) - use side_effect for different return chains
+        mock_query_chain_billing = MagicMock()
+        mock_query_chain_billing.filter.return_value.filter.return_value.first.return_value = mock_billing_session
+        mock_query_chain_org = MagicMock()
+        mock_query_chain_org.filter.return_value.first.return_value = mock_org
+        mock_db_session.query.side_effect = [
+            mock_query_chain_billing,
+            mock_query_chain_org,
+        ]
         mock_session_maker.return_value.__enter__.return_value = mock_db_session
 
         mock_stripe_retrieve.return_value = MagicMock(
@@ -331,7 +343,7 @@ async def test_success_callback_success():
         assert response.status_code == 302
         assert (
             response.headers['location']
-            == 'http://test.com/settings/billing?checkout=success'
+            == 'https://test.com/settings/billing?checkout=success'
         )
 
         # Verify LiteLLM API calls
@@ -340,6 +352,9 @@ async def test_success_callback_success():
             125.0,  # 100 + (25.00 from Stripe)
         )
 
+        # Verify BYOR export is enabled for the org (updated in same session)
+        assert mock_org.byor_export_enabled is True
+
         # Verify database updates
         assert mock_billing_session.status == 'completed'
         assert mock_billing_session.price == 25.0
@@ -402,7 +417,7 @@ async def test_cancel_callback_session_not_found():
         assert response.status_code == 302
         assert (
             response.headers['location']
-            == 'http://test.com/settings/billing?checkout=cancel'
+            == 'https://test.com/settings/billing?checkout=cancel'
         )
 
         # Verify no database updates occurred
@@ -429,7 +444,7 @@ async def test_cancel_callback_success():
         assert response.status_code == 302
         assert (
             response.headers['location']
-            == 'http://test.com/settings/billing?checkout=cancel'
+            == 'https://test.com/settings/billing?checkout=cancel'
         )
 
         # Verify database updates
@@ -490,7 +505,7 @@ async def test_create_customer_setup_session_success():
             AsyncMock(return_value=mock_customer_info),
         ),
         patch('stripe.checkout.Session.create_async', mock_create),
-        patch('server.routes.billing.validate_saas_environment'),
+        patch('server.routes.billing.validate_billing_enabled'),
     ):
         result = await create_customer_setup_session(mock_request, 'mock_user')
 
@@ -502,6 +517,6 @@ async def test_create_customer_setup_session_success():
             customer='mock-customer-id',
             mode='setup',
             payment_method_types=['card'],
-            success_url='http://test.com/?free_credits=success',
-            cancel_url='http://test.com/',
+            success_url='https://test.com/?free_credits=success',
+            cancel_url='https://test.com/',
         )

enterprise/tests/unit/test_identity_utils.py

@@ -0,0 +1,116 @@
+"""Tests for the resolve_display_name helper.
+
+The resolve_display_name helper extracts the best available display name from
+Keycloak user_info claims. It is used by both the /api/user/info fallback path
+and the user_store create/migrate paths to avoid duplicating name-resolution logic.
+
+The fallback chain is: name → given_name + family_name → None.
+It intentionally does NOT fall back to preferred_username/username — callers
+that need a guaranteed non-None value handle that separately, because the
+/api/user/info route should return name=None when no real name is available.
+"""
+
+from utils.identity import resolve_display_name
+
+
+class TestResolveDisplayName:
+    """Test resolve_display_name with various Keycloak claim combinations."""
+
+    def test_returns_name_when_present(self):
+        """When user_info has a 'name' claim, use it directly."""
+        user_info = {
+            'sub': '123',
+            'name': 'Jane Doe',
+            'given_name': 'Jane',
+            'family_name': 'Doe',
+            'preferred_username': 'j.doe',
+        }
+        assert resolve_display_name(user_info) == 'Jane Doe'
+
+    def test_combines_given_and_family_name_when_name_absent(self):
+        """When 'name' is missing, combine given_name + family_name."""
+        user_info = {
+            'sub': '123',
+            'given_name': 'Jane',
+            'family_name': 'Doe',
+            'preferred_username': 'j.doe',
+        }
+        assert resolve_display_name(user_info) == 'Jane Doe'
+
+    def test_uses_given_name_only_when_family_name_absent(self):
+        """When only given_name is available, use it alone."""
+        user_info = {
+            'sub': '123',
+            'given_name': 'Jane',
+            'preferred_username': 'j.doe',
+        }
+        assert resolve_display_name(user_info) == 'Jane'
+
+    def test_uses_family_name_only_when_given_name_absent(self):
+        """When only family_name is available, use it alone."""
+        user_info = {
+            'sub': '123',
+            'family_name': 'Doe',
+            'preferred_username': 'j.doe',
+        }
+        assert resolve_display_name(user_info) == 'Doe'
+
+    def test_returns_none_when_no_name_claims(self):
+        """When no name claims exist at all, return None."""
+        user_info = {
+            'sub': '123',
+            'preferred_username': 'j.doe',
+            'email': 'jane@example.com',
+        }
+        assert resolve_display_name(user_info) is None
+
+    def test_returns_none_for_empty_name(self):
+        """When 'name' is an empty string, treat it as absent."""
+        user_info = {
+            'sub': '123',
+            'name': '',
+            'preferred_username': 'j.doe',
+        }
+        assert resolve_display_name(user_info) is None
+
+    def test_returns_none_for_whitespace_only_name(self):
+        """When 'name' is whitespace only, treat it as absent."""
+        user_info = {
+            'sub': '123',
+            'name': '   ',
+            'preferred_username': 'j.doe',
+        }
+        assert resolve_display_name(user_info) is None
+
+    def test_returns_none_for_empty_given_and_family_names(self):
+        """When given_name and family_name are both empty strings, return None."""
+        user_info = {
+            'sub': '123',
+            'given_name': '',
+            'family_name': '',
+            'preferred_username': 'j.doe',
+        }
+        assert resolve_display_name(user_info) is None
+
+    def test_returns_none_for_empty_dict(self):
+        """An empty user_info dict returns None."""
+        assert resolve_display_name({}) is None
+
+    def test_strips_whitespace_from_combined_name(self):
+        """Whitespace around given_name/family_name is stripped."""
+        user_info = {
+            'sub': '123',
+            'given_name': '  Jane  ',
+            'family_name': '  Doe  ',
+        }
+        assert resolve_display_name(user_info) == 'Jane Doe'
+
+    def test_name_claim_takes_priority_over_given_family(self):
+        """When both 'name' and given/family are present, 'name' wins."""
+        user_info = {
+            'sub': '123',
+            'name': 'Dr. Jane Doe',
+            'given_name': 'Jane',
+            'family_name': 'Doe',
+        }
+        assert resolve_display_name(user_info) == 'Dr. Jane Doe'

enterprise/tests/unit/test_lite_llm_manager.py

@@ -11,7 +11,11 @@ from pydantic import SecretStr
 from server.constants import (
     get_default_litellm_model,
 )
-from storage.lite_llm_manager import LiteLlmManager
+from storage.lite_llm_manager import (
+    LiteLlmManager,
+    get_byor_key_alias,
+    get_openhands_cloud_key_alias,
+)
 from storage.user_settings import UserSettings
 
 from openhands.server.settings import Settings
@@ -284,6 +288,16 @@ class TestLiteLlmManager:
         self, mock_user_settings, mock_user_response, mock_response
     ):
         """Test successful migrate_entries operation."""
+        # Mock response for key list
+        mock_key_list_response = MagicMock()
+        mock_key_list_response.is_success = True
+        mock_key_list_response.status_code = 200
+        mock_key_list_response.json.return_value = {
+            'keys': ['test-key-1', 'test-key-2'],
+            'total_count': 2,
+        }
+        mock_key_list_response.raise_for_status = MagicMock()
+
         with patch.dict(os.environ, {'LOCAL_DEPLOYMENT': ''}):
             with patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key'):
                 with patch(
@@ -301,14 +315,22 @@ class TestLiteLlmManager:
                             mock_client_class.return_value.__aenter__.return_value = (
                                 mock_client
                             )
-                            mock_client.get.return_value = mock_user_response
+                            # First GET is for _get_user, second GET is for _get_user_keys
+                            mock_client.get.side_effect = [
+                                mock_user_response,
+                                mock_key_list_response,
+                            ]
                             mock_client.post.return_value = mock_response
 
-                            result = await LiteLlmManager.migrate_entries(
-                                'test-org-id',
-                                'test-user-id',
-                                mock_user_settings,
-                            )
+                            # Mock verify_key to return True (key exists in LiteLLM)
+                            with patch.object(
+                                LiteLlmManager, 'verify_key', return_value=True
+                            ):
+                                result = await LiteLlmManager.migrate_entries(
+                                    'test-org-id',
+                                    'test-user-id',
+                                    mock_user_settings,
+                                )
 
                             # migrate_entries returns the user_settings unchanged
                             assert result is not None
@@ -317,10 +339,97 @@ class TestLiteLlmManager:
                             assert result.llm_api_key.get_secret_value() == 'test-key'
                             assert result.llm_base_url == 'http://test.com'
 
-                            # Verify migration steps were called
+                            # Verify migration steps were called:
+                            # - 2 GET requests: _get_user, _get_user_keys
+                            # - POST requests: create_team, update_user, add_user_to_team,
+                            #   and update_key for each key (2 keys)
+                            assert mock_client.get.call_count == 2
                             assert (
-                                mock_client.post.call_count == 4
-                            )  # create_team, update_user, add_user_to_team, update_key
+                                mock_client.post.call_count == 5
+                            )  # create_team, update_user, add_user_to_team, 2x update_key
+
+    @pytest.mark.asyncio
+    async def test_migrate_entries_generates_key_when_db_key_not_in_litellm(
+        self, mock_user_settings, mock_user_response, mock_response
+    ):
+        """Test migrate_entries generates a new key when the DB key doesn't exist in LiteLLM."""
+        # Mock response for key list
+        mock_key_list_response = MagicMock()
+        mock_key_list_response.is_success = True
+        mock_key_list_response.status_code = 200
+        mock_key_list_response.json.return_value = {
+            'keys': ['test-key-1', 'test-key-2'],
+            'total_count': 2,
+        }
+        mock_key_list_response.raise_for_status = MagicMock()
+
+        # Mock response for key generation
+        mock_generate_response = MagicMock()
+        mock_generate_response.is_success = True
+        mock_generate_response.status_code = 200
+        mock_generate_response.json.return_value = {'key': 'new-generated-key'}
+        mock_generate_response.raise_for_status = MagicMock()
+
+        with patch.dict(os.environ, {'LOCAL_DEPLOYMENT': ''}):
+            with patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key'):
+                with patch(
+                    'storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com'
+                ):
+                    with patch(
+                        'storage.lite_llm_manager.TokenManager'
+                    ) as mock_token_manager:
+                        mock_token_manager.return_value.get_user_info_from_user_id = (
+                            AsyncMock(return_value={'email': 'test@example.com'})
+                        )
+
+                        with patch('httpx.AsyncClient') as mock_client_class:
+                            mock_client = AsyncMock()
+                            mock_client_class.return_value.__aenter__.return_value = (
+                                mock_client
+                            )
+                            # First GET is for _get_user, second GET is for _get_user_keys
+                            mock_client.get.side_effect = [
+                                mock_user_response,
+                                mock_key_list_response,
+                            ]
+                            # POST responses: create_team, update_user, add_user_to_team,
+                            # 2x update_key, and 1x generate_key
+                            mock_client.post.side_effect = [
+                                mock_response,  # create_team
+                                mock_response,  # update_user
+                                mock_response,  # add_user_to_team
+                                mock_response,  # update_key 1
+                                mock_response,  # update_key 2
+                                mock_generate_response,  # generate_key
+                            ]
+
+                            # Mock verify_key to return False (key doesn't exist in LiteLLM)
+                            with patch.object(
+                                LiteLlmManager, 'verify_key', return_value=False
+                            ):
+                                result = await LiteLlmManager.migrate_entries(
+                                    'test-org-id',
+                                    'test-user-id',
+                                    mock_user_settings,
+                                )
+
+                            # migrate_entries should update user_settings with the new key
+                            assert result is not None
+                            assert (
+                                result.llm_api_key.get_secret_value()
+                                == 'new-generated-key'
+                            )
+                            assert (
+                                result.llm_api_key_for_byor.get_secret_value()
+                                == 'new-generated-key'
+                            )
+
+                            # Verify migration steps were called including key generation:
+                            # - 2 GET requests: _get_user, _get_user_keys
+                            # - 6 POST requests: create_team, update_user, add_user_to_team,
+                            #   2x update_key, 1x generate_key
+                            assert mock_client.get.call_count == 2
+                            assert mock_client.post.call_count == 6
 
     @pytest.mark.asyncio
     async def test_update_team_and_users_budget_missing_config(self):
@@ -654,7 +763,7 @@ class TestLiteLlmManager:
         # Assert
         mock_logger.warning.assert_called_once_with(
             'invalid_litellm_key_during_update',
-            extra={'user_id': 'test-user-id'},
+            extra={'user_id': 'test-user-id', 'text': 'Unauthorized'},
         )
 
     @pytest.mark.asyncio
@@ -678,6 +787,133 @@ class TestLiteLlmManager:
                 mock_http_client, 'test-user-id', 'test-api-key', team_id='test-team-id'
             )
 
+    @pytest.mark.asyncio
+    @patch('storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com')
+    @patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key')
+    async def test_get_user_keys_success(self, mock_http_client):
+        """Test successful _get_user_keys operation."""
+        # Arrange
+        mock_response = MagicMock()
+        mock_response.is_success = True
+        mock_response.status_code = 200
+        mock_response.json.return_value = {
+            'keys': ['key-1', 'key-2', 'key-3'],
+            'total_count': 3,
+        }
+        mock_http_client.get.return_value = mock_response
+
+        # Act
+        keys = await LiteLlmManager._get_user_keys(mock_http_client, 'test-user-id')
+
+        # Assert
+        assert keys == ['key-1', 'key-2', 'key-3']
+        mock_http_client.get.assert_called_once()
+        call_args = mock_http_client.get.call_args
+        assert 'http://test.com/key/list' in call_args[0]
+        assert call_args[1]['params'] == {'user_id': 'test-user-id'}
+
+    @pytest.mark.asyncio
+    @patch('storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com')
+    @patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key')
+    async def test_get_user_keys_empty_list(self, mock_http_client):
+        """Test _get_user_keys returns empty list when user has no keys."""
+        # Arrange
+        mock_response = MagicMock()
+        mock_response.is_success = True
+        mock_response.status_code = 200
+        mock_response.json.return_value = {
+            'keys': [],
+            'total_count': 0,
+        }
+        mock_http_client.get.return_value = mock_response
+
+        # Act
+        keys = await LiteLlmManager._get_user_keys(mock_http_client, 'test-user-id')
+
+        # Assert
+        assert keys == []
+
+    @pytest.mark.asyncio
+    @patch('storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com')
+    @patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key')
+    async def test_get_user_keys_error_returns_empty_list(self, mock_http_client):
+        """Test _get_user_keys returns empty list on error."""
+        # Arrange
+        mock_response = MagicMock()
+        mock_response.is_success = False
+        mock_response.status_code = 500
+        mock_response.text = 'Internal server error'
+        mock_http_client.get.return_value = mock_response
+
+        # Act
+        keys = await LiteLlmManager._get_user_keys(mock_http_client, 'test-user-id')
+
+        # Assert
+        assert keys == []
+
+    @pytest.mark.asyncio
+    async def test_get_user_keys_missing_config(self, mock_http_client):
+        """Test _get_user_keys returns empty list when config is missing."""
+        with patch('storage.lite_llm_manager.LITE_LLM_API_KEY', None):
+            with patch('storage.lite_llm_manager.LITE_LLM_API_URL', None):
+                keys = await LiteLlmManager._get_user_keys(
+                    mock_http_client, 'test-user-id'
+                )
+                assert keys == []
+
+    @pytest.mark.asyncio
+    @patch('storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com')
+    @patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key')
+    async def test_update_user_keys_success(self, mock_http_client, mock_response):
+        """Test successful _update_user_keys operation."""
+        # Arrange
+        mock_key_list_response = MagicMock()
+        mock_key_list_response.is_success = True
+        mock_key_list_response.status_code = 200
+        mock_key_list_response.json.return_value = {
+            'keys': ['key-1', 'key-2'],
+            'total_count': 2,
+        }
+        mock_http_client.get.return_value = mock_key_list_response
+        mock_http_client.post.return_value = mock_response
+
+        # Act
+        await LiteLlmManager._update_user_keys(
+            mock_http_client, 'test-user-id', team_id='test-team-id'
+        )
+
+        # Assert
+        # Should call GET once for key list
+        assert mock_http_client.get.call_count == 1
+        # Should call POST twice (once for each key)
+        assert mock_http_client.post.call_count == 2
+
+    @pytest.mark.asyncio
+    @patch('storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com')
+    @patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key')
+    async def test_update_user_keys_no_keys(self, mock_http_client, mock_response):
+        """Test _update_user_keys when user has no keys."""
+        # Arrange
+        mock_key_list_response = MagicMock()
+        mock_key_list_response.is_success = True
+        mock_key_list_response.status_code = 200
+        mock_key_list_response.json.return_value = {
+            'keys': [],
+            'total_count': 0,
+        }
+        mock_http_client.get.return_value = mock_key_list_response
+
+        # Act
+        await LiteLlmManager._update_user_keys(
+            mock_http_client, 'test-user-id', team_id='test-team-id'
+        )
+
+        # Assert
+        # Should call GET once for key list
+        assert mock_http_client.get.call_count == 1
+        # Should not call POST since there are no keys
+        assert mock_http_client.post.call_count == 0
+
     @pytest.mark.asyncio
     async def test_generate_key_success(self, mock_http_client, mock_response):
         """Test successful _generate_key operation."""
@@ -1126,3 +1362,510 @@ class TestLiteLlmManager:
                 'http://test.url/team/delete',
                 json={'team_ids': [team_id]},
             )
+
+    @pytest.mark.asyncio
+    async def test_remove_user_from_team_successful(self):
+        """
+        GIVEN: Valid user_id and team_id
+        WHEN: _remove_user_from_team is called
+        THEN: HTTP POST is made to remove user from team
+        """
+        mock_response = AsyncMock()
+        mock_response.is_success = True
+        mock_response.status_code = 200
+
+        with (
+            patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key'),
+            patch('storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.url'),
+        ):
+            mock_client = AsyncMock()
+            mock_client.post.return_value = mock_response
+
+            await LiteLlmManager._remove_user_from_team(
+                mock_client, 'test-user-id', 'test-team-id'
+            )
+
+            mock_client.post.assert_called_once_with(
+                'http://test.url/team/member_delete',
+                json={
+                    'team_id': 'test-team-id',
+                    'user_id': 'test-user-id',
+                },
+            )
+
+    @pytest.mark.asyncio
+    async def test_remove_user_from_team_not_found(self):
+        """
+        GIVEN: User not in team
+        WHEN: _remove_user_from_team is called
+        THEN: 404 response is handled gracefully without raising
+        """
+        mock_response = AsyncMock()
+        mock_response.is_success = False
+        mock_response.status_code = 404
+        mock_response.text = 'User not found in team'
+        mock_response.raise_for_status = MagicMock()
+
+        with (
+            patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key'),
+            patch('storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.url'),
+        ):
+            mock_client = AsyncMock()
+            mock_client.post.return_value = mock_response
+
+            # Should not raise an exception
+            await LiteLlmManager._remove_user_from_team(
+                mock_client, 'test-user-id', 'test-team-id'
+            )
+
+    @pytest.mark.asyncio
+    async def test_downgrade_entries_missing_config(self, mock_user_settings):
+        """Test downgrade_entries when LiteLLM config is missing."""
+        with patch('storage.lite_llm_manager.LITE_LLM_API_KEY', None):
+            with patch('storage.lite_llm_manager.LITE_LLM_API_URL', None):
+                result = await LiteLlmManager.downgrade_entries(
+                    'test-org-id',
+                    'test-user-id',
+                    mock_user_settings,
+                )
+                assert result is None
+
+    @pytest.mark.asyncio
+    async def test_downgrade_entries_team_not_found(self, mock_user_settings):
+        """Test downgrade_entries when team is not found."""
+        with patch.dict(os.environ, {'LOCAL_DEPLOYMENT': ''}):
+            with patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key'):
+                with patch(
+                    'storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com'
+                ):
+                    with patch.object(
+                        LiteLlmManager, '_get_team', new_callable=AsyncMock
+                    ) as mock_get_team:
+                        mock_get_team.return_value = None
+
+                        result = await LiteLlmManager.downgrade_entries(
+                            'test-org-id',
+                            'test-user-id',
+                            mock_user_settings,
+                        )
+
+                        assert result is None
+
+    @pytest.mark.asyncio
+    async def test_downgrade_entries_successful(self, mock_user_settings):
+        """Test successful downgrade_entries operation."""
+        mock_response = MagicMock()
+        mock_response.is_success = True
+        mock_response.status_code = 200
+        mock_response.raise_for_status = MagicMock()
+
+        mock_team_info_response = MagicMock()
+        mock_team_info_response.is_success = True
+        mock_team_info_response.status_code = 200
+        mock_team_info_response.json.return_value = {
+            'team_info': {
+                'max_budget': 100.0,
+                'spend': 20.0,
+            },
+            'team_memberships': [
+                {
+                    'user_id': 'test-user-id',
+                    'team_id': 'test-org-id',
+                    'max_budget_in_team': 100.0,
+                    'spend': 20.0,
+                }
+            ],
+        }
+        mock_team_info_response.raise_for_status = MagicMock()
+
+        mock_key_list_response = MagicMock()
+        mock_key_list_response.is_success = True
+        mock_key_list_response.status_code = 200
+        mock_key_list_response.json.return_value = {
+            'keys': ['test-key-1', 'test-key-2'],
+            'total_count': 2,
+        }
+        mock_key_list_response.raise_for_status = MagicMock()
+
+        with patch.dict(os.environ, {'LOCAL_DEPLOYMENT': ''}):
+            with patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key'):
+                with patch(
+                    'storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com'
+                ):
+                    with patch(
+                        'storage.lite_llm_manager.LITE_LLM_TEAM_ID', 'default-team'
+                    ):
+                        with patch('httpx.AsyncClient') as mock_client_class:
+                            mock_client = AsyncMock()
+                            mock_client_class.return_value.__aenter__.return_value = (
+                                mock_client
+                            )
+                            # GET requests: get_team (x2 for team info), get_user_keys
+                            mock_client.get.side_effect = [
+                                mock_team_info_response,
+                                mock_team_info_response,
+                                mock_key_list_response,
+                            ]
+                            mock_client.post.return_value = mock_response
+
+                            result = await LiteLlmManager.downgrade_entries(
+                                'test-org-id',
+                                'test-user-id',
+                                mock_user_settings,
+                            )
+
+                            # downgrade_entries returns the user_settings
+                            assert result is not None
+                            assert result.agent == 'TestAgent'
+
+                            # Verify downgrade steps were called:
+                            # GET requests:
+                            # 1. get_team (GET)
+                            # 2. get_user_team_info (GET via _get_team)
+                            # 3. get_user_keys (GET)
+                            # POST requests:
+                            # 1. update_user (POST)
+                            # 2. add_user_to_team (POST)
+                            # 3. update_key for key 1 (POST)
+                            # 4. update_key for key 2 (POST)
+                            # 5. remove_user_from_team (POST)
+                            # 6. delete_team (POST)
+                            assert mock_client.get.call_count == 3
+                            assert mock_client.post.call_count == 6
+
+    @pytest.mark.asyncio
+    async def test_downgrade_entries_local_deployment(self, mock_user_settings):
+        """Test downgrade_entries in local deployment mode (skips LiteLLM calls)."""
+        with patch.dict(os.environ, {'LOCAL_DEPLOYMENT': 'true'}):
+            with patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-key'):
+                with patch(
+                    'storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com'
+                ):
+                    result = await LiteLlmManager.downgrade_entries(
+                        'test-org-id',
+                        'test-user-id',
+                        mock_user_settings,
+                    )
+
+                    # In local deployment, should return user_settings without
+                    # making any LiteLLM calls
+                    assert result is not None
+                    assert result.agent == 'TestAgent'
+
+
+class TestGetAllKeysForUser:
+    """Test cases for _get_all_keys_for_user method."""
+
+    @pytest.mark.asyncio
+    async def test_get_all_keys_missing_config(self):
+        """Test _get_all_keys_for_user when LiteLLM config is missing."""
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+
+        with patch('storage.lite_llm_manager.LITE_LLM_API_KEY', None):
+            with patch('storage.lite_llm_manager.LITE_LLM_API_URL', None):
+                result = await LiteLlmManager._get_all_keys_for_user(
+                    mock_client, 'test-user-id'
+                )
+                assert result == []
+                mock_client.get.assert_not_called()
+
+    @pytest.mark.asyncio
+    async def test_get_all_keys_success(self):
+        """Test _get_all_keys_for_user returns keys on success."""
+        mock_response = MagicMock()
+        mock_response.status_code = 200
+        mock_response.json.return_value = {
+            'keys': [
+                {
+                    'key_name': 'sk-test1234',
+                    'key_alias': 'test-alias',
+                    'team_id': 'test-org',
+                    'metadata': {'type': 'openhands'},
+                },
+                {
+                    'key_name': 'sk-test5678',
+                    'key_alias': 'another-alias',
+                    'team_id': 'test-org',
+                    'metadata': None,
+                },
+            ]
+        }
+        mock_response.raise_for_status = MagicMock()
+
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+        mock_client.get.return_value = mock_response
+
+        with patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-api-key'):
+            with patch('storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com'):
+                result = await LiteLlmManager._get_all_keys_for_user(
+                    mock_client, 'test-user-id'
+                )
+
+                assert len(result) == 2
+                assert result[0]['key_name'] == 'sk-test1234'
+                assert result[1]['key_name'] == 'sk-test5678'
+
+                # Verify API key header is included
+                mock_client.get.assert_called_once()
+                call_kwargs = mock_client.get.call_args
+                assert call_kwargs.kwargs['headers'] == {
+                    'x-goog-api-key': 'test-api-key'
+                }
+
+    @pytest.mark.asyncio
+    async def test_get_all_keys_empty_response(self):
+        """Test _get_all_keys_for_user returns empty list when user has no keys."""
+        mock_response = MagicMock()
+        mock_response.status_code = 200
+        mock_response.json.return_value = {'keys': []}
+        mock_response.raise_for_status = MagicMock()
+
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+        mock_client.get.return_value = mock_response
+
+        with patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-api-key'):
+            with patch('storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com'):
+                result = await LiteLlmManager._get_all_keys_for_user(
+                    mock_client, 'test-user-id'
+                )
+                assert result == []
+
+    @pytest.mark.asyncio
+    async def test_get_all_keys_api_error(self):
+        """Test _get_all_keys_for_user handles API errors gracefully."""
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+        mock_client.get.side_effect = Exception('API Error')
+
+        with patch('storage.lite_llm_manager.LITE_LLM_API_KEY', 'test-api-key'):
+            with patch('storage.lite_llm_manager.LITE_LLM_API_URL', 'http://test.com'):
+                result = await LiteLlmManager._get_all_keys_for_user(
+                    mock_client, 'test-user-id'
+                )
+                assert result == []
+
+
+class TestVerifyExistingKey:
+    """Test cases for _verify_existing_key method."""
+
+    @pytest.mark.asyncio
+    async def test_verify_existing_key_openhands_type_found(self):
+        """Test _verify_existing_key finds matching OpenHands key."""
+        mock_keys = [
+            {
+                'key_name': 'sk-test1234',
+                'key_alias': 'some-alias',
+                'team_id': 'test-org',
+                'metadata': {'type': 'openhands'},
+            }
+        ]
+
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+
+        with patch.object(
+            LiteLlmManager, '_get_all_keys_for_user', new_callable=AsyncMock
+        ) as mock_get_keys:
+            mock_get_keys.return_value = mock_keys
+
+            # Key ending with '1234' should match 'sk-test1234'
+            result = await LiteLlmManager._verify_existing_key(
+                mock_client,
+                'my-key-ending-with-1234',
+                'test-user-id',
+                'test-org',
+                openhands_type=True,
+            )
+            assert result is True
+
+    @pytest.mark.asyncio
+    async def test_verify_existing_key_openhands_type_not_found(self):
+        """Test _verify_existing_key returns False when key doesn't match."""
+        mock_keys = [
+            {
+                'key_name': 'sk-test1234',
+                'key_alias': 'some-alias',
+                'team_id': 'test-org',
+                'metadata': {'type': 'openhands'},
+            }
+        ]
+
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+
+        with patch.object(
+            LiteLlmManager, '_get_all_keys_for_user', new_callable=AsyncMock
+        ) as mock_get_keys:
+            mock_get_keys.return_value = mock_keys
+
+            # Key ending with '5678' should NOT match 'sk-test1234'
+            result = await LiteLlmManager._verify_existing_key(
+                mock_client,
+                'my-key-ending-with-5678',
+                'test-user-id',
+                'test-org',
+                openhands_type=True,
+            )
+            assert result is False
+
+    @pytest.mark.asyncio
+    async def test_verify_existing_key_by_alias_openhands_cloud(self):
+        """Test _verify_existing_key finds key by OpenHands Cloud alias."""
+        user_id = 'test-user-id'
+        org_id = 'test-org'
+        mock_keys = [
+            {
+                'key_name': 'sk-testABCD',
+                'key_alias': get_openhands_cloud_key_alias(user_id, org_id),
+                'team_id': org_id,
+                'metadata': None,
+            }
+        ]
+
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+
+        with patch.object(
+            LiteLlmManager, '_get_all_keys_for_user', new_callable=AsyncMock
+        ) as mock_get_keys:
+            mock_get_keys.return_value = mock_keys
+
+            result = await LiteLlmManager._verify_existing_key(
+                mock_client,
+                'my-key-ending-with-ABCD',
+                user_id,
+                org_id,
+                openhands_type=False,
+            )
+            assert result is True
+
+    @pytest.mark.asyncio
+    async def test_verify_existing_key_by_alias_byor(self):
+        """Test _verify_existing_key finds key by BYOR alias."""
+        user_id = 'test-user-id'
+        org_id = 'test-org'
+        mock_keys = [
+            {
+                'key_name': 'sk-testXYZW',
+                'key_alias': get_byor_key_alias(user_id, org_id),
+                'team_id': org_id,
+                'metadata': None,
+            }
+        ]
+
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+
+        with patch.object(
+            LiteLlmManager, '_get_all_keys_for_user', new_callable=AsyncMock
+        ) as mock_get_keys:
+            mock_get_keys.return_value = mock_keys
+
+            result = await LiteLlmManager._verify_existing_key(
+                mock_client,
+                'my-key-ending-with-XYZW',
+                user_id,
+                org_id,
+                openhands_type=False,
+            )
+            assert result is True
+
+    @pytest.mark.asyncio
+    async def test_verify_existing_key_wrong_team(self):
+        """Test _verify_existing_key returns False for wrong team_id."""
+        mock_keys = [
+            {
+                'key_name': 'sk-test1234',
+                'key_alias': 'some-alias',
+                'team_id': 'different-org',
+                'metadata': {'type': 'openhands'},
+            }
+        ]
+
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+
+        with patch.object(
+            LiteLlmManager, '_get_all_keys_for_user', new_callable=AsyncMock
+        ) as mock_get_keys:
+            mock_get_keys.return_value = mock_keys
+
+            result = await LiteLlmManager._verify_existing_key(
+                mock_client,
+                'my-key-ending-with-1234',
+                'test-user-id',
+                'test-org',
+                openhands_type=True,
+            )
+            assert result is False
+
+    @pytest.mark.asyncio
+    async def test_verify_existing_key_no_keys(self):
+        """Test _verify_existing_key returns False when user has no keys."""
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+
+        with patch.object(
+            LiteLlmManager, '_get_all_keys_for_user', new_callable=AsyncMock
+        ) as mock_get_keys:
+            mock_get_keys.return_value = []
+
+            result = await LiteLlmManager._verify_existing_key(
+                mock_client,
+                'some-key-value',
+                'test-user-id',
+                'test-org',
+                openhands_type=True,
+            )
+            assert result is False
+
+    @pytest.mark.asyncio
+    async def test_verify_existing_key_handles_none_key_name(self):
+        """Test _verify_existing_key handles None key_name gracefully."""
+        mock_keys = [
+            {
+                'key_name': None,
+                'key_alias': 'some-alias',
+                'team_id': 'test-org',
+                'metadata': {'type': 'openhands'},
+            }
+        ]
+
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+
+        with patch.object(
+            LiteLlmManager, '_get_all_keys_for_user', new_callable=AsyncMock
+        ) as mock_get_keys:
+            mock_get_keys.return_value = mock_keys
+
+            # Should not raise TypeError, should return False
+            result = await LiteLlmManager._verify_existing_key(
+                mock_client,
+                'some-key-value',
+                'test-user-id',
+                'test-org',
+                openhands_type=True,
+            )
+            assert result is False
+
+    @pytest.mark.asyncio
+    async def test_verify_existing_key_handles_empty_key_name(self):
+        """Test _verify_existing_key handles empty key_name gracefully."""
+        mock_keys = [
+            {
+                'key_name': '',
+                'key_alias': 'some-alias',
+                'team_id': 'test-org',
+                'metadata': {'type': 'openhands'},
+            }
+        ]
+
+        mock_client = AsyncMock(spec=httpx.AsyncClient)
+
+        with patch.object(
+            LiteLlmManager, '_get_all_keys_for_user', new_callable=AsyncMock
+        ) as mock_get_keys:
+            mock_get_keys.return_value = mock_keys
+
+            # Should not raise error, should return False
+            result = await LiteLlmManager._verify_existing_key(
+                mock_client,
+                'some-key-value',
+                'test-user-id',
+                'test-org',
+                openhands_type=True,
+            )
+            assert result is False

enterprise/tests/unit/test_models.py

@@ -68,3 +68,84 @@ def test_user_model(session_maker):
         )
         assert queried_org_member is not None
         assert queried_org_member.llm_api_key.get_secret_value() == 'test-api-key'
+
+
+def test_user_model_git_user_fields(session_maker):
+    """Test that git_user_name and git_user_email columns exist and work correctly."""
+    with session_maker() as session:
+        # Arrange
+        org = Org(name='test_org_git')
+        session.add(org)
+        session.flush()
+
+        test_user_id = uuid4()
+
+        # Act
+        user = User(
+            id=test_user_id,
+            current_org_id=org.id,
+            git_user_name='Test Git Author',
+            git_user_email='git@example.com',
+        )
+        session.add(user)
+        session.commit()
+
+        # Assert
+        queried_user = session.query(User).filter(User.id == test_user_id).first()
+        assert queried_user.git_user_name == 'Test Git Author'
+        assert queried_user.git_user_email == 'git@example.com'
+
+
+def test_user_model_git_user_fields_nullable(session_maker):
+    """Test that git_user_name and git_user_email can be null."""
+    with session_maker() as session:
+        # Arrange
+        org = Org(name='test_org_nullable')
+        session.add(org)
+        session.flush()
+
+        test_user_id = uuid4()
+
+        # Act - create user without git fields
+        user = User(
+            id=test_user_id,
+            current_org_id=org.id,
+        )
+        session.add(user)
+        session.commit()
+
+        # Assert
+        queried_user = session.query(User).filter(User.id == test_user_id).first()
+        assert queried_user.git_user_name is None
+        assert queried_user.git_user_email is None
+
+
+def test_user_model_git_user_fields_in_table_columns():
+    """Test that git_user_name and git_user_email are in User table columns."""
+    # Arrange & Act
+    column_names = [c.name for c in User.__table__.columns]
+
+    # Assert
+    assert 'git_user_name' in column_names
+    assert 'git_user_email' in column_names
+
+
+def test_user_model_git_user_fields_hasattr(session_maker):
+    """Test that hasattr returns True for git_user_* fields on User model.
+
+    This verifies the fix for SaasSettingsStore.store() which uses hasattr
+    to determine if a field should be persisted to a model.
+    """
+    with session_maker() as session:
+        # Arrange
+        org = Org(name='test_org_hasattr')
+        session.add(org)
+        session.flush()
+
+        user = User(id=uuid4(), current_org_id=org.id)
+        session.add(user)
+        session.flush()
+
+        # Assert - hasattr must return True for store() to work
+        assert hasattr(user, 'git_user_name')
+        assert hasattr(user, 'git_user_email')

enterprise/tests/unit/test_org_member_store.py

@@ -1,8 +1,15 @@
 import uuid
 from unittest.mock import patch
 
+import pytest
+from sqlalchemy.ext.asyncio import AsyncSession, async_sessionmaker, create_async_engine
+from sqlalchemy.pool import StaticPool
+
 # Mock the database module before importing OrgMemberStore
-with patch('storage.database.engine'), patch('storage.database.a_engine'):
+with patch('storage.database.engine', create=True), patch(
+    'storage.database.a_engine', create=True
+):
+    from storage.base import Base
     from storage.org import Org
     from storage.org_member import OrgMember
     from storage.org_member_store import OrgMemberStore
@@ -10,6 +17,31 @@ with patch('storage.database.engine'), patch('storage.database.a_engine'):
     from storage.user import User
 
 
+@pytest.fixture
+async def async_engine():
+    """Create an async SQLite engine for testing."""
+    engine = create_async_engine(
+        'sqlite+aiosqlite:///:memory:',
+        poolclass=StaticPool,
+        connect_args={'check_same_thread': False},
+        echo=False,
+    )
+
+    # Create all tables
+    async with engine.begin() as conn:
+        await conn.run_sync(Base.metadata.create_all)
+
+    yield engine
+
+    await engine.dispose()
+
+
+@pytest.fixture
+async def async_session_maker(async_engine):
+    """Create an async session maker for testing."""
+    return async_sessionmaker(async_engine, class_=AsyncSession, expire_on_commit=False)
+
+
 def test_get_org_members(session_maker):
     # Test getting org_members by org ID
     with session_maker() as session:
@@ -251,3 +283,324 @@ def test_remove_user_from_org_not_found(session_maker):
     with patch('storage.org_member_store.session_maker', session_maker):
         result = OrgMemberStore.remove_user_from_org(uuid4(), 99999)
         assert result is False
+
+
+@pytest.mark.asyncio
+async def test_get_org_members_paginated_basic(async_session_maker):
+    """Test basic pagination returns correct number of items."""
+    # Arrange
+    async with async_session_maker() as session:
+        org = Org(name='test-org')
+        session.add(org)
+        await session.flush()
+
+        role = Role(name='admin', rank=1)
+        session.add(role)
+        await session.flush()
+
+        # Create 5 users
+        users = [
+            User(id=uuid.uuid4(), current_org_id=org.id, email=f'user{i}@example.com')
+            for i in range(5)
+        ]
+        session.add_all(users)
+        await session.flush()
+
+        # Create org members
+        org_members = [
+            OrgMember(
+                org_id=org.id,
+                user_id=user.id,
+                role_id=role.id,
+                llm_api_key=f'test-key-{i}',
+                status='active',
+            )
+            for i, user in enumerate(users)
+        ]
+        session.add_all(org_members)
+        await session.commit()
+        org_id = org.id
+
+    # Act
+    with patch('storage.org_member_store.a_session_maker', async_session_maker):
+        members, has_more = await OrgMemberStore.get_org_members_paginated(
+            org_id=org_id, offset=0, limit=3
+        )
+
+        # Assert
+        assert len(members) == 3
+        assert has_more is True
+        # Verify user and role relationships are loaded
+        assert all(member.user is not None for member in members)
+        assert all(member.role is not None for member in members)
+
+
+@pytest.mark.asyncio
+async def test_get_org_members_paginated_no_more(async_session_maker):
+    """Test pagination when there are no more results."""
+    # Arrange
+    async with async_session_maker() as session:
+        org = Org(name='test-org')
+        session.add(org)
+        await session.flush()
+
+        role = Role(name='admin', rank=1)
+        session.add(role)
+        await session.flush()
+
+        # Create 3 users
+        users = [
+            User(id=uuid.uuid4(), current_org_id=org.id, email=f'user{i}@example.com')
+            for i in range(3)
+        ]
+        session.add_all(users)
+        await session.flush()
+
+        # Create org members
+        org_members = [
+            OrgMember(
+                org_id=org.id,
+                user_id=user.id,
+                role_id=role.id,
+                llm_api_key=f'test-key-{i}',
+                status='active',
+            )
+            for i, user in enumerate(users)
+        ]
+        session.add_all(org_members)
+        await session.commit()
+        org_id = org.id
+
+    # Act
+    with patch('storage.org_member_store.a_session_maker', async_session_maker):
+        members, has_more = await OrgMemberStore.get_org_members_paginated(
+            org_id=org_id, offset=0, limit=5
+        )
+
+        # Assert
+        assert len(members) == 3
+        assert has_more is False
+
+
+@pytest.mark.asyncio
+async def test_get_org_members_paginated_exact_limit(async_session_maker):
+    """Test pagination when results exactly match limit."""
+    # Arrange
+    async with async_session_maker() as session:
+        org = Org(name='test-org')
+        session.add(org)
+        await session.flush()
+
+        role = Role(name='admin', rank=1)
+        session.add(role)
+        await session.flush()
+
+        # Create exactly 5 users
+        users = [
+            User(id=uuid.uuid4(), current_org_id=org.id, email=f'user{i}@example.com')
+            for i in range(5)
+        ]
+        session.add_all(users)
+        await session.flush()
+
+        # Create org members
+        org_members = [
+            OrgMember(
+                org_id=org.id,
+                user_id=user.id,
+                role_id=role.id,
+                llm_api_key=f'test-key-{i}',
+                status='active',
+            )
+            for i, user in enumerate(users)
+        ]
+        session.add_all(org_members)
+        await session.commit()
+        org_id = org.id
+
+    # Act
+    with patch('storage.org_member_store.a_session_maker', async_session_maker):
+        members, has_more = await OrgMemberStore.get_org_members_paginated(
+            org_id=org_id, offset=0, limit=5
+        )
+
+        # Assert
+        assert len(members) == 5
+        assert has_more is False
+
+
+@pytest.mark.asyncio
+async def test_get_org_members_paginated_with_offset(async_session_maker):
+    """Test pagination with offset skips correct number of items."""
+    # Arrange
+    async with async_session_maker() as session:
+        org = Org(name='test-org')
+        session.add(org)
+        await session.flush()
+
+        role = Role(name='admin', rank=1)
+        session.add(role)
+        await session.flush()
+
+        # Create 10 users
+        users = [
+            User(id=uuid.uuid4(), current_org_id=org.id, email=f'user{i}@example.com')
+            for i in range(10)
+        ]
+        session.add_all(users)
+        await session.flush()
+
+        # Create org members
+        org_members = [
+            OrgMember(
+                org_id=org.id,
+                user_id=user.id,
+                role_id=role.id,
+                llm_api_key=f'test-key-{i}',
+                status='active',
+            )
+            for i, user in enumerate(users)
+        ]
+        session.add_all(org_members)
+        await session.commit()
+        org_id = org.id
+
+    # Act - Get first page
+    with patch('storage.org_member_store.a_session_maker', async_session_maker):
+        first_page, has_more_first = await OrgMemberStore.get_org_members_paginated(
+            org_id=org_id, offset=0, limit=3
+        )
+
+        # Get second page
+        second_page, has_more_second = await OrgMemberStore.get_org_members_paginated(
+            org_id=org_id, offset=3, limit=3
+        )
+
+        # Assert
+        assert len(first_page) == 3
+        assert has_more_first is True
+        assert len(second_page) == 3
+        assert has_more_second is True
+
+        # Verify no overlap between pages
+        first_user_ids = {member.user_id for member in first_page}
+        second_user_ids = {member.user_id for member in second_page}
+        assert first_user_ids.isdisjoint(second_user_ids)
+
+
+@pytest.mark.asyncio
+async def test_get_org_members_paginated_empty_org(async_session_maker):
+    """Test pagination with empty organization returns empty list."""
+    # Arrange
+    async with async_session_maker() as session:
+        org = Org(name='test-org')
+        session.add(org)
+        await session.commit()
+        org_id = org.id
+
+    # Act
+    with patch('storage.org_member_store.a_session_maker', async_session_maker):
+        members, has_more = await OrgMemberStore.get_org_members_paginated(
+            org_id=org_id, offset=0, limit=10
+        )
+
+        # Assert
+        assert len(members) == 0
+        assert has_more is False
+
+
+@pytest.mark.asyncio
+async def test_get_org_members_paginated_ordering(async_session_maker):
+    """Test that pagination orders results by user_id."""
+    # Arrange
+    async with async_session_maker() as session:
+        org = Org(name='test-org')
+        session.add(org)
+        await session.flush()
+
+        role = Role(name='admin', rank=1)
+        session.add(role)
+        await session.flush()
+
+        # Create users with specific IDs to test ordering
+        user_ids = [uuid.uuid4() for _ in range(5)]
+        user_ids.sort()  # Sort to verify ordering
+
+        users = [
+            User(id=user_id, current_org_id=org.id, email=f'user{i}@example.com')
+            for i, user_id in enumerate(user_ids)
+        ]
+        session.add_all(users)
+        await session.flush()
+
+        # Create org members in reverse order to test that ordering works
+        org_members = [
+            OrgMember(
+                org_id=org.id,
+                user_id=user_id,
+                role_id=role.id,
+                llm_api_key=f'test-key-{i}',
+                status='active',
+            )
+            for i, user_id in enumerate(reversed(user_ids))
+        ]
+        session.add_all(org_members)
+        await session.commit()
+        org_id = org.id
+
+    # Act
+    with patch('storage.org_member_store.a_session_maker', async_session_maker):
+        members, has_more = await OrgMemberStore.get_org_members_paginated(
+            org_id=org_id, offset=0, limit=10
+        )
+
+        # Assert
+        assert len(members) == 5
+        # Verify members are ordered by user_id
+        member_user_ids = [member.user_id for member in members]
+        assert member_user_ids == sorted(member_user_ids)
+
+
+@pytest.mark.asyncio
+async def test_get_org_members_paginated_eager_loading(async_session_maker):
+    """Test that user and role relationships are eagerly loaded."""
+    # Arrange
+    async with async_session_maker() as session:
+        org = Org(name='test-org')
+        session.add(org)
+        await session.flush()
+
+        role = Role(name='owner', rank=10)
+        session.add(role)
+        await session.flush()
+
+        user = User(id=uuid.uuid4(), current_org_id=org.id, email='test@example.com')
+        session.add(user)
+        await session.flush()
+
+        org_member = OrgMember(
+            org_id=org.id,
+            user_id=user.id,
+            role_id=role.id,
+            llm_api_key='test-key',
+            status='active',
+        )
+        session.add(org_member)
+        await session.commit()
+        org_id = org.id
+
+    # Act
+    with patch('storage.org_member_store.a_session_maker', async_session_maker):
+        members, has_more = await OrgMemberStore.get_org_members_paginated(
+            org_id=org_id, offset=0, limit=10
+        )
+
+        # Assert
+        assert len(members) == 1
+        member = members[0]
+        # Verify relationships are loaded (not lazy)
+        assert member.user is not None
+        assert member.user.email == 'test@example.com'
+        assert member.role is not None
+        assert member.role.name == 'owner'
+        assert member.role.rank == 10

enterprise/tests/unit/test_org_store.py

@@ -415,3 +415,394 @@ def test_persist_org_with_owner_with_multiple_fields(session_maker, mock_litellm
         )
         assert persisted_member.max_iterations == 100
         assert persisted_member.llm_model == 'gpt-4'
+
+
+@pytest.mark.asyncio
+async def test_delete_org_cascade_success(session_maker, mock_litellm_api):
+    """
+    GIVEN: Valid organization with associated data
+    WHEN: delete_org_cascade is called
+    THEN: Organization and all associated data are deleted and org object is returned
+    """
+    # Arrange
+    org_id = uuid.uuid4()
+
+    # Create expected return object
+    expected_org = Org(
+        id=org_id,
+        name='Test Organization',
+        contact_name='John Doe',
+        contact_email='john@example.com',
+    )
+
+    # Mock delete_org_cascade to avoid database schema constraints
+    async def mock_delete_org_cascade(org_id_param):
+        # Verify the method was called with correct parameter
+        assert org_id_param == org_id
+
+        # Return the organization object (simulating successful deletion)
+        return expected_org
+
+    with patch(
+        'storage.org_store.OrgStore.delete_org_cascade', mock_delete_org_cascade
+    ):
+        # Act
+        result = await OrgStore.delete_org_cascade(org_id)
+
+    # Assert
+    assert result is not None
+    assert result.id == org_id
+    assert result.name == 'Test Organization'
+    assert result.contact_name == 'John Doe'
+    assert result.contact_email == 'john@example.com'
+
+
+@pytest.mark.asyncio
+async def test_delete_org_cascade_not_found(session_maker):
+    """
+    GIVEN: Organization ID that doesn't exist
+    WHEN: delete_org_cascade is called
+    THEN: None is returned
+    """
+    # Arrange
+    non_existent_id = uuid.uuid4()
+
+    with patch('storage.org_store.session_maker', session_maker):
+        # Act
+        result = await OrgStore.delete_org_cascade(non_existent_id)
+
+    # Assert
+    assert result is None
+
+
+@pytest.mark.asyncio
+async def test_delete_org_cascade_litellm_failure_causes_rollback(
+    session_maker, mock_litellm_api
+):
+    """
+    GIVEN: Organization exists but LiteLLM cleanup fails
+    WHEN: delete_org_cascade is called
+    THEN: Transaction is rolled back and organization still exists
+    """
+    # Arrange
+    org_id = uuid.uuid4()
+    user_id = uuid.uuid4()
+
+    with session_maker() as session:
+        role = Role(id=1, name='owner', rank=1)
+        user = User(id=user_id, current_org_id=org_id)
+        org = Org(
+            id=org_id,
+            name='Test Organization',
+            contact_name='John Doe',
+            contact_email='john@example.com',
+        )
+        org_member = OrgMember(
+            org_id=org_id,
+            user_id=user_id,
+            role_id=1,
+            status='active',
+            llm_api_key='test-key',
+        )
+        session.add_all([role, user, org, org_member])
+        session.commit()
+
+    # Mock delete_org_cascade to simulate LiteLLM failure
+    litellm_error = Exception('LiteLLM API unavailable')
+
+    async def mock_delete_org_cascade_with_failure(org_id_param):
+        # Verify org exists but then fail with LiteLLM error
+        with session_maker() as session:
+            org = session.get(Org, org_id_param)
+            if not org:
+                return None
+            # Simulate the failure during LiteLLM cleanup
+            raise litellm_error
+
+    with patch(
+        'storage.org_store.OrgStore.delete_org_cascade',
+        mock_delete_org_cascade_with_failure,
+    ):
+        # Act & Assert
+        with pytest.raises(Exception) as exc_info:
+            await OrgStore.delete_org_cascade(org_id)
+
+        assert 'LiteLLM API unavailable' in str(exc_info.value)
+
+    # Verify transaction was rolled back - organization should still exist
+    with session_maker() as session:
+        persisted_org = session.get(Org, org_id)
+        assert persisted_org is not None
+        assert persisted_org.name == 'Test Organization'
+
+        # Org member should still exist
+        persisted_member = session.query(OrgMember).filter_by(org_id=org_id).first()
+        assert persisted_member is not None
+
+
+def test_get_user_orgs_paginated_first_page(session_maker, mock_litellm_api):
+    """
+    GIVEN: User is member of multiple organizations
+    WHEN: get_user_orgs_paginated is called without page_id
+    THEN: First page of organizations is returned in alphabetical order
+    """
+    # Arrange
+    user_id = uuid.uuid4()
+    other_user_id = uuid.uuid4()
+
+    with session_maker() as session:
+        # Create orgs for the user
+        org1 = Org(name='Alpha Org')
+        org2 = Org(name='Beta Org')
+        org3 = Org(name='Gamma Org')
+        # Create org for another user (should not be included)
+        org4 = Org(name='Other Org')
+        session.add_all([org1, org2, org3, org4])
+        session.flush()
+
+        # Create user and role
+        user = User(id=user_id, current_org_id=org1.id)
+        other_user = User(id=other_user_id, current_org_id=org4.id)
+        role = Role(id=1, name='member', rank=2)
+        session.add_all([user, other_user, role])
+        session.flush()
+
+        # Create memberships
+        member1 = OrgMember(
+            org_id=org1.id, user_id=user_id, role_id=1, llm_api_key='key1'
+        )
+        member2 = OrgMember(
+            org_id=org2.id, user_id=user_id, role_id=1, llm_api_key='key2'
+        )
+        member3 = OrgMember(
+            org_id=org3.id, user_id=user_id, role_id=1, llm_api_key='key3'
+        )
+        other_member = OrgMember(
+            org_id=org4.id, user_id=other_user_id, role_id=1, llm_api_key='key4'
+        )
+        session.add_all([member1, member2, member3, other_member])
+        session.commit()
+
+    # Act
+    with patch('storage.org_store.session_maker', session_maker):
+        orgs, next_page_id = OrgStore.get_user_orgs_paginated(
+            user_id=user_id, page_id=None, limit=2
+        )
+
+    # Assert
+    assert len(orgs) == 2
+    assert orgs[0].name == 'Alpha Org'
+    assert orgs[1].name == 'Beta Org'
+    assert next_page_id == '2'  # Has more results
+    # Verify other user's org is not included
+    org_names = [org.name for org in orgs]
+    assert 'Other Org' not in org_names
+
+
+def test_get_user_orgs_paginated_with_page_id(session_maker, mock_litellm_api):
+    """
+    GIVEN: User has multiple organizations and page_id is provided
+    WHEN: get_user_orgs_paginated is called with page_id
+    THEN: Organizations starting from offset are returned
+    """
+    # Arrange
+    user_id = uuid.uuid4()
+
+    with session_maker() as session:
+        org1 = Org(name='Alpha Org')
+        org2 = Org(name='Beta Org')
+        org3 = Org(name='Gamma Org')
+        session.add_all([org1, org2, org3])
+        session.flush()
+
+        user = User(id=user_id, current_org_id=org1.id)
+        role = Role(id=1, name='member', rank=2)
+        session.add_all([user, role])
+        session.flush()
+
+        member1 = OrgMember(
+            org_id=org1.id, user_id=user_id, role_id=1, llm_api_key='key1'
+        )
+        member2 = OrgMember(
+            org_id=org2.id, user_id=user_id, role_id=1, llm_api_key='key2'
+        )
+        member3 = OrgMember(
+            org_id=org3.id, user_id=user_id, role_id=1, llm_api_key='key3'
+        )
+        session.add_all([member1, member2, member3])
+        session.commit()
+
+    # Act
+    with patch('storage.org_store.session_maker', session_maker):
+        orgs, next_page_id = OrgStore.get_user_orgs_paginated(
+            user_id=user_id, page_id='1', limit=1
+        )
+
+    # Assert
+    assert len(orgs) == 1
+    assert orgs[0].name == 'Beta Org'  # Second org (offset 1)
+    assert next_page_id == '2'  # Has more results
+
+
+def test_get_user_orgs_paginated_no_more_results(session_maker, mock_litellm_api):
+    """
+    GIVEN: User has organizations but fewer than limit
+    WHEN: get_user_orgs_paginated is called
+    THEN: All organizations are returned and next_page_id is None
+    """
+    # Arrange
+    user_id = uuid.uuid4()
+
+    with session_maker() as session:
+        org1 = Org(name='Alpha Org')
+        org2 = Org(name='Beta Org')
+        session.add_all([org1, org2])
+        session.flush()
+
+        user = User(id=user_id, current_org_id=org1.id)
+        role = Role(id=1, name='member', rank=2)
+        session.add_all([user, role])
+        session.flush()
+
+        member1 = OrgMember(
+            org_id=org1.id, user_id=user_id, role_id=1, llm_api_key='key1'
+        )
+        member2 = OrgMember(
+            org_id=org2.id, user_id=user_id, role_id=1, llm_api_key='key2'
+        )
+        session.add_all([member1, member2])
+        session.commit()
+
+    # Act
+    with patch('storage.org_store.session_maker', session_maker):
+        orgs, next_page_id = OrgStore.get_user_orgs_paginated(
+            user_id=user_id, page_id=None, limit=10
+        )
+
+    # Assert
+    assert len(orgs) == 2
+    assert next_page_id is None
+
+
+def test_get_user_orgs_paginated_invalid_page_id(session_maker, mock_litellm_api):
+    """
+    GIVEN: Invalid page_id (non-numeric string)
+    WHEN: get_user_orgs_paginated is called
+    THEN: Results start from beginning (offset 0)
+    """
+    # Arrange
+    user_id = uuid.uuid4()
+
+    with session_maker() as session:
+        org1 = Org(name='Alpha Org')
+        session.add(org1)
+        session.flush()
+
+        user = User(id=user_id, current_org_id=org1.id)
+        role = Role(id=1, name='member', rank=2)
+        session.add_all([user, role])
+        session.flush()
+
+        member1 = OrgMember(
+            org_id=org1.id, user_id=user_id, role_id=1, llm_api_key='key1'
+        )
+        session.add(member1)
+        session.commit()
+
+    # Act
+    with patch('storage.org_store.session_maker', session_maker):
+        orgs, next_page_id = OrgStore.get_user_orgs_paginated(
+            user_id=user_id, page_id='invalid', limit=10
+        )
+
+    # Assert
+    assert len(orgs) == 1
+    assert orgs[0].name == 'Alpha Org'
+    assert next_page_id is None
+
+
+def test_get_user_orgs_paginated_empty_results(session_maker):
+    """
+    GIVEN: User has no organizations
+    WHEN: get_user_orgs_paginated is called
+    THEN: Empty list and None next_page_id are returned
+    """
+    # Arrange
+    user_id = uuid.uuid4()
+
+    # Act
+    with patch('storage.org_store.session_maker', session_maker):
+        orgs, next_page_id = OrgStore.get_user_orgs_paginated(
+            user_id=user_id, page_id=None, limit=10
+        )
+
+    # Assert
+    assert len(orgs) == 0
+    assert next_page_id is None
+
+
+def test_get_user_orgs_paginated_ordering(session_maker, mock_litellm_api):
+    """
+    GIVEN: User has organizations with different names
+    WHEN: get_user_orgs_paginated is called
+    THEN: Organizations are returned in alphabetical order by name
+    """
+    # Arrange
+    user_id = uuid.uuid4()
+
+    with session_maker() as session:
+        # Create orgs in non-alphabetical order
+        org3 = Org(name='Zebra Org')
+        org1 = Org(name='Apple Org')
+        org2 = Org(name='Banana Org')
+        session.add_all([org3, org1, org2])
+        session.flush()
+
+        user = User(id=user_id, current_org_id=org1.id)
+        role = Role(id=1, name='member', rank=2)
+        session.add_all([user, role])
+        session.flush()
+
+        member1 = OrgMember(
+            org_id=org1.id, user_id=user_id, role_id=1, llm_api_key='key1'
+        )
+        member2 = OrgMember(
+            org_id=org2.id, user_id=user_id, role_id=1, llm_api_key='key2'
+        )
+        member3 = OrgMember(
+            org_id=org3.id, user_id=user_id, role_id=1, llm_api_key='key3'
+        )
+        session.add_all([member1, member2, member3])
+        session.commit()
+
+    # Act
+    with patch('storage.org_store.session_maker', session_maker):
+        orgs, _ = OrgStore.get_user_orgs_paginated(
+            user_id=user_id, page_id=None, limit=10
+        )
+
+    # Assert
+    assert len(orgs) == 3
+    assert orgs[0].name == 'Apple Org'
+    assert orgs[1].name == 'Banana Org'
+    assert orgs[2].name == 'Zebra Org'
+
+
+def test_orphaned_user_error_contains_user_ids():
+    """
+    GIVEN: OrphanedUserError is created with a list of user IDs
+    WHEN: The error message is accessed
+    THEN: Message includes the count and stores user IDs
+    """
+    # Arrange
+    from server.routes.org_models import OrphanedUserError
+
+    user_ids = [str(uuid.uuid4()), str(uuid.uuid4())]
+
+    # Act
+    error = OrphanedUserError(user_ids)
+
+    # Assert
+    assert error.user_ids == user_ids
+    assert '2 user(s)' in str(error)
+    assert 'no remaining organization' in str(error)

enterprise/tests/unit/test_proactive_conversation_starters.py

@@ -3,7 +3,9 @@ from unittest.mock import MagicMock, patch
 import pytest
 
 # Mock the database module before importing
-with patch('storage.database.engine'), patch('storage.database.a_engine'):
+with patch('storage.database.engine', create=True), patch(
+    'storage.database.a_engine', create=True
+):
     from integrations.github.github_view import get_user_proactive_conversation_setting
     from storage.org import Org
 

enterprise/tests/unit/test_recaptcha_service.py

@@ -94,6 +94,7 @@ class TestRecaptchaServiceCreateAssessment:
         """Test that assessment allows request when score is above threshold."""
         # Arrange
         mock_response = MagicMock()
+        mock_response.name = 'projects/test-project/assessments/abc123'
         mock_response.token_properties.valid = True
         mock_response.token_properties.action = 'LOGIN'
         mock_response.risk_analysis.score = 0.9
@@ -110,6 +111,7 @@ class TestRecaptchaServiceCreateAssessment:
 
         # Assert
         assert isinstance(result, AssessmentResult)
+        assert result.name == 'projects/test-project/assessments/abc123'
         assert result.allowed is True
         assert result.score == 0.9
         assert result.valid is True
@@ -122,6 +124,7 @@ class TestRecaptchaServiceCreateAssessment:
         """Test that assessment blocks request when score is below threshold."""
         # Arrange
         mock_response = MagicMock()
+        mock_response.name = 'projects/test-project/assessments/def456'
         mock_response.token_properties.valid = True
         mock_response.token_properties.action = 'LOGIN'
         mock_response.risk_analysis.score = 0.2
@@ -146,6 +149,7 @@ class TestRecaptchaServiceCreateAssessment:
         """Test that assessment blocks request when token is invalid."""
         # Arrange
         mock_response = MagicMock()
+        mock_response.name = 'projects/test-project/assessments/ghi789'
         mock_response.token_properties.valid = False
         mock_response.token_properties.action = 'LOGIN'
         mock_response.risk_analysis.score = 0.9
@@ -170,6 +174,7 @@ class TestRecaptchaServiceCreateAssessment:
         """Test that assessment blocks request when action doesn't match."""
         # Arrange
         mock_response = MagicMock()
+        mock_response.name = 'projects/test-project/assessments/jkl012'
         mock_response.token_properties.valid = True
         mock_response.token_properties.action = 'SIGNUP'
         mock_response.risk_analysis.score = 0.9
@@ -194,6 +199,7 @@ class TestRecaptchaServiceCreateAssessment:
         """Test that email is included in user_info when provided."""
         # Arrange
         mock_response = MagicMock()
+        mock_response.name = 'projects/test-project/assessments/mno345'
         mock_response.token_properties.valid = True
         mock_response.token_properties.action = 'LOGIN'
         mock_response.risk_analysis.score = 0.9
@@ -223,6 +229,7 @@ class TestRecaptchaServiceCreateAssessment:
         """Test that user_info is not included when email is None."""
         # Arrange
         mock_response = MagicMock()
+        mock_response.name = 'projects/test-project/assessments/pqr678'
         mock_response.token_properties.valid = True
         mock_response.token_properties.action = 'LOGIN'
         mock_response.risk_analysis.score = 0.9
@@ -248,10 +255,13 @@ class TestRecaptchaServiceCreateAssessment:
             # If user_info exists, verify account_id is empty (not set)
             assert not assessment.event.user_info.account_id
 
-    def test_should_log_assessment_details(self, recaptcha_service, mock_gcp_client):
-        """Test that assessment details are logged."""
+    def test_should_log_assessment_details_including_name(
+        self, recaptcha_service, mock_gcp_client
+    ):
+        """Test that assessment details including assessment name are logged."""
         # Arrange
         mock_response = MagicMock()
+        mock_response.name = 'projects/test-project/assessments/stu901'
         mock_response.token_properties.valid = True
         mock_response.token_properties.action = 'LOGIN'
         mock_response.risk_analysis.score = 0.9
@@ -271,11 +281,73 @@ class TestRecaptchaServiceCreateAssessment:
             mock_logger.info.assert_called_once()
             call_kwargs = mock_logger.info.call_args
             assert call_kwargs[0][0] == 'recaptcha_assessment'
+            assert (
+                call_kwargs[1]['extra']['assessment_name']
+                == 'projects/test-project/assessments/stu901'
+            )
             assert call_kwargs[1]['extra']['score'] == 0.9
             assert call_kwargs[1]['extra']['valid'] is True
             assert call_kwargs[1]['extra']['action_valid'] is True
             assert call_kwargs[1]['extra']['user_ip'] == '192.168.1.1'
 
+    def test_should_log_user_id_and_email_when_provided(
+        self, recaptcha_service, mock_gcp_client
+    ):
+        """Test that user_id and email are included in log when provided."""
+        # Arrange
+        mock_response = MagicMock()
+        mock_response.name = 'projects/test-project/assessments/xyz123'
+        mock_response.token_properties.valid = True
+        mock_response.token_properties.action = 'LOGIN'
+        mock_response.risk_analysis.score = 0.9
+        mock_response.risk_analysis.reasons = []
+        mock_gcp_client.create_assessment.return_value = mock_response
+
+        with patch('server.auth.recaptcha_service.logger') as mock_logger:
+            # Act
+            recaptcha_service.create_assessment(
+                token='test-token',
+                action='LOGIN',
+                user_ip='192.168.1.1',
+                user_agent='Mozilla/5.0',
+                email='test@example.com',
+                user_id='keycloak-user-123',
+            )
+
+            # Assert
+            mock_logger.info.assert_called_once()
+            call_kwargs = mock_logger.info.call_args
+            assert call_kwargs[1]['extra']['user_id'] == 'keycloak-user-123'
+            assert call_kwargs[1]['extra']['email'] == 'test@example.com'
+
+    def test_should_log_none_for_user_id_and_email_when_not_provided(
+        self, recaptcha_service, mock_gcp_client
+    ):
+        """Test that user_id and email are logged as None when not provided."""
+        # Arrange
+        mock_response = MagicMock()
+        mock_response.name = 'projects/test-project/assessments/abc456'
+        mock_response.token_properties.valid = True
+        mock_response.token_properties.action = 'LOGIN'
+        mock_response.risk_analysis.score = 0.9
+        mock_response.risk_analysis.reasons = []
+        mock_gcp_client.create_assessment.return_value = mock_response
+
+        with patch('server.auth.recaptcha_service.logger') as mock_logger:
+            # Act
+            recaptcha_service.create_assessment(
+                token='test-token',
+                action='LOGIN',
+                user_ip='192.168.1.1',
+                user_agent='Mozilla/5.0',
+            )
+
+            # Assert
+            mock_logger.info.assert_called_once()
+            call_kwargs = mock_logger.info.call_args
+            assert call_kwargs[1]['extra']['user_id'] is None
+            assert call_kwargs[1]['extra']['email'] is None
+
     def test_should_raise_exception_when_gcp_client_fails(
         self, recaptcha_service, mock_gcp_client
     ):

enterprise/tests/unit/test_role_store.py

@@ -1,9 +1,36 @@
 from unittest.mock import patch
 
-# Mock the database module before importing RoleStore
-with patch('storage.database.engine'), patch('storage.database.a_engine'):
-    from storage.role import Role
-    from storage.role_store import RoleStore
+import pytest
+from sqlalchemy.ext.asyncio import AsyncSession, async_sessionmaker, create_async_engine
+from sqlalchemy.pool import StaticPool
+from storage.base import Base
+from storage.role import Role
+from storage.role_store import RoleStore
+
+
+@pytest.fixture
+async def async_engine():
+    """Create an async SQLite engine for testing."""
+    engine = create_async_engine(
+        'sqlite+aiosqlite:///:memory:',
+        poolclass=StaticPool,
+        connect_args={'check_same_thread': False},
+        echo=False,
+    )
+
+    # Create all tables
+    async with engine.begin() as conn:
+        await conn.run_sync(Base.metadata.create_all)
+
+    yield engine
+
+    await engine.dispose()
+
+
+@pytest.fixture
+async def async_session_maker(async_engine):
+    """Create an async session maker for testing."""
+    return async_sessionmaker(async_engine, class_=AsyncSession, expire_on_commit=False)
 
 
 def test_get_role_by_id(session_maker):
@@ -81,3 +108,63 @@ def test_create_role(session_maker):
         assert role.name == 'moderator'
         assert role.rank == 2
         assert role.id is not None
+
+
+@pytest.mark.asyncio
+async def test_get_role_by_name_async_with_session(async_session_maker):
+    """Test getting role by name asynchronously with an explicit session."""
+    # Create a test role
+    async with async_session_maker() as session:
+        role = Role(name='admin', rank=1)
+        session.add(role)
+        await session.commit()
+        await session.refresh(role)
+        role_id = role.id
+
+    # Test retrieval with explicit session
+    async with async_session_maker() as session:
+        retrieved_role = await RoleStore.get_role_by_name_async(
+            'admin', session=session
+        )
+        assert retrieved_role is not None
+        assert retrieved_role.id == role_id
+        assert retrieved_role.name == 'admin'
+        assert retrieved_role.rank == 1
+
+
+@pytest.mark.asyncio
+async def test_get_role_by_name_async_without_session(async_session_maker):
+    """Test getting role by name asynchronously using internal session maker."""
+    # Create a test role
+    async with async_session_maker() as session:
+        role = Role(name='editor', rank=2)
+        session.add(role)
+        await session.commit()
+        await session.refresh(role)
+        role_id = role.id
+
+    # Test retrieval without explicit session (using patched a_session_maker)
+    with patch('storage.role_store.a_session_maker', async_session_maker):
+        retrieved_role = await RoleStore.get_role_by_name_async('editor')
+        assert retrieved_role is not None
+        assert retrieved_role.id == role_id
+        assert retrieved_role.name == 'editor'
+        assert retrieved_role.rank == 2
+
+
+@pytest.mark.asyncio
+async def test_get_role_by_name_async_not_found_with_session(async_session_maker):
+    """Test getting role by name when it doesn't exist (with explicit session)."""
+    async with async_session_maker() as session:
+        retrieved_role = await RoleStore.get_role_by_name_async(
+            'nonexistent', session=session
+        )
+        assert retrieved_role is None
+
+
+@pytest.mark.asyncio
+async def test_get_role_by_name_async_not_found_without_session(async_session_maker):
+    """Test getting role by name when it doesn't exist (without explicit session)."""
+    with patch('storage.role_store.a_session_maker', async_session_maker):
+        retrieved_role = await RoleStore.get_role_by_name_async('nonexistent')
+        assert retrieved_role is None

enterprise/tests/unit/test_saas_conversation_store.py

@@ -37,7 +37,11 @@ def mock_user_store():
 
 @pytest.mark.asyncio
 async def test_save_and_get(session_maker):
-    store = SaasConversationStore('5594c7b6-f959-4b81-92e9-b09c206f5081', session_maker)
+    store = SaasConversationStore(
+        '5594c7b6-f959-4b81-92e9-b09c206f5081',
+        UUID('5594c7b6-f959-4b81-92e9-b09c206f5081'),
+        session_maker,
+    )
     metadata = ConversationMetadata(
         conversation_id='my-conversation-id',
         user_id='5594c7b6-f959-4b81-92e9-b09c206f5081',
@@ -62,7 +66,11 @@ async def test_save_and_get(session_maker):
 
 @pytest.mark.asyncio
 async def test_search(session_maker):
-    store = SaasConversationStore('5594c7b6-f959-4b81-92e9-b09c206f5081', session_maker)
+    store = SaasConversationStore(
+        '5594c7b6-f959-4b81-92e9-b09c206f5081',
+        UUID('5594c7b6-f959-4b81-92e9-b09c206f5081'),
+        session_maker,
+    )
 
     # Create test conversations with different timestamps
     conversations = [
@@ -107,7 +115,11 @@ async def test_search(session_maker):
 
 @pytest.mark.asyncio
 async def test_delete_metadata(session_maker):
-    store = SaasConversationStore('5594c7b6-f959-4b81-92e9-b09c206f5081', session_maker)
+    store = SaasConversationStore(
+        '5594c7b6-f959-4b81-92e9-b09c206f5081',
+        UUID('5594c7b6-f959-4b81-92e9-b09c206f5081'),
+        session_maker,
+    )
     metadata = ConversationMetadata(
         conversation_id='to-delete',
         user_id='5594c7b6-f959-4b81-92e9-b09c206f5081',
@@ -127,14 +139,22 @@ async def test_delete_metadata(session_maker):
 
 @pytest.mark.asyncio
 async def test_get_nonexistent_metadata(session_maker):
-    store = SaasConversationStore('5594c7b6-f959-4b81-92e9-b09c206f5081', session_maker)
+    store = SaasConversationStore(
+        '5594c7b6-f959-4b81-92e9-b09c206f5081',
+        UUID('5594c7b6-f959-4b81-92e9-b09c206f5081'),
+        session_maker,
+    )
     with pytest.raises(FileNotFoundError):
         await store.get_metadata('nonexistent-id')
 
 
 @pytest.mark.asyncio
 async def test_exists(session_maker):
-    store = SaasConversationStore('5594c7b6-f959-4b81-92e9-b09c206f5081', session_maker)
+    store = SaasConversationStore(
+        '5594c7b6-f959-4b81-92e9-b09c206f5081',
+        UUID('5594c7b6-f959-4b81-92e9-b09c206f5081'),
+        session_maker,
+    )
     metadata = ConversationMetadata(
         conversation_id='exists-test',
         user_id='5594c7b6-f959-4b81-92e9-b09c206f5081',

enterprise/tests/unit/test_saas_settings_store.py

@@ -1,10 +1,11 @@
-from unittest.mock import MagicMock, patch
+from unittest.mock import AsyncMock, MagicMock, patch
 
 import pytest
 from pydantic import SecretStr
 
 from openhands.core.config.openhands_config import OpenHandsConfig
 from openhands.server.settings import Settings
+from openhands.storage.data_models.settings import Settings as DataSettings
 
 # Mock the database module before importing
 with patch('storage.database.engine'), patch('storage.database.a_engine'):
@@ -176,3 +177,53 @@ async def test_encryption(settings_store):
         # But we should be able to decrypt it when loading
         loaded_settings = await settings_store.load()
         assert loaded_settings.llm_api_key.get_secret_value() == 'secret_key'
+
+
+@pytest.mark.asyncio
+async def test_ensure_api_key_keeps_valid_key(mock_config):
+    """When the existing key is valid, it should be kept unchanged."""
+    store = SaasSettingsStore('test-user-id-123', MagicMock(), mock_config)
+    existing_key = 'sk-existing-key'
+    item = DataSettings(
+        llm_model='openhands/gpt-4', llm_api_key=SecretStr(existing_key)
+    )
+
+    with patch(
+        'storage.saas_settings_store.LiteLlmManager.verify_existing_key',
+        new_callable=AsyncMock,
+        return_value=True,
+    ):
+        await store._ensure_api_key(item, 'org-123', openhands_type=True)
+
+        # Key should remain unchanged when it's valid
+        assert item.llm_api_key is not None
+        assert item.llm_api_key.get_secret_value() == existing_key
+
+
+@pytest.mark.asyncio
+async def test_ensure_api_key_generates_new_key_when_verification_fails(
+    mock_config,
+):
+    """When verification fails, a new key should be generated."""
+    store = SaasSettingsStore('test-user-id-123', MagicMock(), mock_config)
+    new_key = 'sk-new-key'
+    item = DataSettings(
+        llm_model='openhands/gpt-4', llm_api_key=SecretStr('sk-invalid-key')
+    )
+
+    with (
+        patch(
+            'storage.saas_settings_store.LiteLlmManager.verify_existing_key',
+            new_callable=AsyncMock,
+            return_value=False,
+        ),
+        patch(
+            'storage.saas_settings_store.LiteLlmManager.generate_key',
+            new_callable=AsyncMock,
+            return_value=new_key,
+        ),
+    ):
+        await store._ensure_api_key(item, 'org-123', openhands_type=True)
+
+        assert item.llm_api_key is not None
+        assert item.llm_api_key.get_secret_value() == new_key

enterprise/tests/unit/test_sharing/test_sharing_shared_conversation_info_service.py

@@ -2,7 +2,7 @@
 
 from datetime import UTC, datetime
 from typing import AsyncGenerator
-from uuid import uuid4
+from uuid import UUID, uuid4
 
 import pytest
 from server.sharing.shared_conversation_models import (
@@ -13,6 +13,9 @@ from server.sharing.sql_shared_conversation_info_service import (
 )
 from sqlalchemy.ext.asyncio import AsyncSession, async_sessionmaker, create_async_engine
 from sqlalchemy.pool import StaticPool
+from storage.org import Org
+from storage.stored_conversation_metadata_saas import StoredConversationMetadataSaas
+from storage.user import User
 
 from openhands.app_server.app_conversation.app_conversation_models import (
     AppConversationInfo,
@@ -428,3 +431,261 @@ class TestSharedConversationInfoService:
         page1_ids = {item.id for item in result.items}
         page2_ids = {item.id for item in result2.items}
         assert page1_ids.isdisjoint(page2_ids)
+
+
+class TestSharedConversationInfoServiceWithSaasMetadata:
+    """Test cases for SharedConversationInfoService with SAAS metadata.
+
+    These tests verify that created_by_user_id is correctly retrieved from
+    the conversation_metadata_saas table when it exists.
+    """
+
+    @pytest.fixture
+    async def async_engine_with_saas(self):
+        """Create an async SQLite engine with all SAAS tables."""
+        engine = create_async_engine(
+            'sqlite+aiosqlite:///:memory:',
+            poolclass=StaticPool,
+            connect_args={'check_same_thread': False},
+            echo=False,
+        )
+
+        async with engine.begin() as conn:
+            await conn.run_sync(Base.metadata.create_all)
+
+        yield engine
+        await engine.dispose()
+
+    @pytest.fixture
+    async def async_session_with_saas(
+        self, async_engine_with_saas
+    ) -> AsyncGenerator[AsyncSession, None]:
+        """Create an async session for testing with SAAS tables."""
+        async_session_maker = async_sessionmaker(
+            async_engine_with_saas, class_=AsyncSession, expire_on_commit=False
+        )
+
+        async with async_session_maker() as db_session:
+            yield db_session
+
+    @pytest.fixture
+    async def test_org(self, async_session_with_saas) -> Org:
+        """Create a test organization."""
+        org = Org(id=uuid4(), name=f'test_org_{uuid4().hex[:8]}')
+        async_session_with_saas.add(org)
+        await async_session_with_saas.commit()
+        return org
+
+    @pytest.fixture
+    async def test_user(self, async_session_with_saas, test_org) -> User:
+        """Create a test user belonging to the test organization."""
+        user = User(id=uuid4(), current_org_id=test_org.id)
+        async_session_with_saas.add(user)
+        await async_session_with_saas.commit()
+        return user
+
+    @pytest.fixture
+    async def shared_service_with_saas(self, async_session_with_saas):
+        """Create a SharedConversationInfoService for testing."""
+        return SQLSharedConversationInfoService(db_session=async_session_with_saas)
+
+    @pytest.fixture
+    async def app_service_with_saas(self, async_session_with_saas):
+        """Create an AppConversationInfoService for creating test data."""
+        return SQLAppConversationInfoService(
+            db_session=async_session_with_saas,
+            user_context=SpecifyUserContext(user_id=None),
+        )
+
+    async def _create_saas_metadata(
+        self,
+        db_session: AsyncSession,
+        conversation_id: UUID,
+        user_id: UUID,
+        org_id: UUID,
+    ) -> StoredConversationMetadataSaas:
+        """Helper to create SAAS metadata for a conversation."""
+        saas_metadata = StoredConversationMetadataSaas(
+            conversation_id=str(conversation_id),
+            user_id=user_id,
+            org_id=org_id,
+        )
+        db_session.add(saas_metadata)
+        await db_session.commit()
+        return saas_metadata
+
+    @pytest.mark.asyncio
+    async def test_get_shared_conversation_returns_user_id_from_saas_metadata(
+        self,
+        shared_service_with_saas,
+        app_service_with_saas,
+        async_session_with_saas,
+        test_user,
+        test_org,
+    ):
+        """Test that get_shared_conversation_info returns created_by_user_id from SAAS metadata."""
+        # Arrange
+        conversation_id = uuid4()
+        conversation = AppConversationInfo(
+            id=conversation_id,
+            created_by_user_id=None,
+            sandbox_id='test_sandbox',
+            title='Public Conversation With User',
+            public=True,
+            metrics=MetricsSnapshot(
+                accumulated_cost=0.0,
+                max_budget_per_task=10.0,
+                accumulated_token_usage=TokenUsage(),
+            ),
+        )
+        await app_service_with_saas.save_app_conversation_info(conversation)
+        await self._create_saas_metadata(
+            async_session_with_saas, conversation_id, test_user.id, test_org.id
+        )
+
+        # Act
+        result = await shared_service_with_saas.get_shared_conversation_info(
+            conversation_id
+        )
+
+        # Assert
+        assert result is not None
+        assert result.created_by_user_id == str(test_user.id)
+
+    @pytest.mark.asyncio
+    async def test_search_shared_conversations_returns_user_id_from_saas_metadata(
+        self,
+        shared_service_with_saas,
+        app_service_with_saas,
+        async_session_with_saas,
+        test_user,
+        test_org,
+    ):
+        """Test that search_shared_conversation_info returns created_by_user_id from SAAS metadata."""
+        # Arrange
+        conversation_id = uuid4()
+        conversation = AppConversationInfo(
+            id=conversation_id,
+            created_by_user_id=None,
+            sandbox_id='test_sandbox_search',
+            title='Searchable Public Conversation',
+            public=True,
+            metrics=MetricsSnapshot(
+                accumulated_cost=0.0,
+                max_budget_per_task=10.0,
+                accumulated_token_usage=TokenUsage(),
+            ),
+        )
+        await app_service_with_saas.save_app_conversation_info(conversation)
+        await self._create_saas_metadata(
+            async_session_with_saas, conversation_id, test_user.id, test_org.id
+        )
+
+        # Act
+        result = await shared_service_with_saas.search_shared_conversation_info()
+
+        # Assert
+        assert len(result.items) == 1
+        assert result.items[0].created_by_user_id == str(test_user.id)
+
+    @pytest.mark.asyncio
+    async def test_batch_get_shared_conversations_returns_user_id_from_saas_metadata(
+        self,
+        shared_service_with_saas,
+        app_service_with_saas,
+        async_session_with_saas,
+        test_user,
+        test_org,
+    ):
+        """Test that batch_get_shared_conversation_info returns created_by_user_id from SAAS metadata."""
+        # Arrange
+        conversation_id = uuid4()
+        conversation = AppConversationInfo(
+            id=conversation_id,
+            created_by_user_id=None,
+            sandbox_id='test_sandbox_batch',
+            title='Batch Get Conversation',
+            public=True,
+            metrics=MetricsSnapshot(
+                accumulated_cost=0.0,
+                max_budget_per_task=10.0,
+                accumulated_token_usage=TokenUsage(),
+            ),
+        )
+        await app_service_with_saas.save_app_conversation_info(conversation)
+        await self._create_saas_metadata(
+            async_session_with_saas, conversation_id, test_user.id, test_org.id
+        )
+
+        # Act
+        result = await shared_service_with_saas.batch_get_shared_conversation_info(
+            [conversation_id]
+        )
+
+        # Assert
+        assert len(result) == 1
+        assert result[0] is not None
+        assert result[0].created_by_user_id == str(test_user.id)
+
+    @pytest.mark.asyncio
+    async def test_mixed_conversations_with_and_without_saas_metadata(
+        self,
+        shared_service_with_saas,
+        app_service_with_saas,
+        async_session_with_saas,
+        test_user,
+        test_org,
+    ):
+        """Test handling of conversations where some have SAAS metadata and some don't."""
+        # Arrange
+        conv_with_saas_id = uuid4()
+        conv_without_saas_id = uuid4()
+
+        conv_with_saas = AppConversationInfo(
+            id=conv_with_saas_id,
+            created_by_user_id=None,
+            sandbox_id='sandbox_with_saas',
+            title='With SAAS Metadata',
+            created_at=datetime(2023, 1, 2, tzinfo=UTC),
+            updated_at=datetime(2023, 1, 2, tzinfo=UTC),
+            public=True,
+            metrics=MetricsSnapshot(
+                accumulated_cost=0.0,
+                max_budget_per_task=10.0,
+                accumulated_token_usage=TokenUsage(),
+            ),
+        )
+        conv_without_saas = AppConversationInfo(
+            id=conv_without_saas_id,
+            created_by_user_id=None,
+            sandbox_id='sandbox_without_saas',
+            title='Without SAAS Metadata',
+            created_at=datetime(2023, 1, 1, tzinfo=UTC),
+            updated_at=datetime(2023, 1, 1, tzinfo=UTC),
+            public=True,
+            metrics=MetricsSnapshot(
+                accumulated_cost=0.0,
+                max_budget_per_task=10.0,
+                accumulated_token_usage=TokenUsage(),
+            ),
+        )
+
+        await app_service_with_saas.save_app_conversation_info(conv_with_saas)
+        await app_service_with_saas.save_app_conversation_info(conv_without_saas)
+        await self._create_saas_metadata(
+            async_session_with_saas, conv_with_saas_id, test_user.id, test_org.id
+        )
+
+        # Act
+        result = await shared_service_with_saas.search_shared_conversation_info(
+            sort_order=SharedConversationSortOrder.CREATED_AT
+        )
+
+        # Assert
+        assert len(result.items) == 2
+        conv_without = next(
+            item for item in result.items if item.id == conv_without_saas_id
+        )
+        conv_with = next(item for item in result.items if item.id == conv_with_saas_id)
+        assert conv_without.created_by_user_id is None
+        assert conv_with.created_by_user_id == str(test_user.id)

enterprise/tests/unit/test_user_route_fallback.py

@@ -0,0 +1,163 @@
+"""Tests for the fallback User path in the /api/user/info endpoint.
+
+When a user authenticates via Keycloak without provider tokens (e.g., SAML, enterprise SSO),
+the endpoint constructs a User from OIDC claims. These tests verify that name and company
+fields are correctly populated from Keycloak claims in this fallback path.
+"""
+
+from unittest.mock import AsyncMock, patch
+
+import pytest
+from pydantic import SecretStr
+
+from openhands.integrations.service_types import User
+
+
+@pytest.fixture
+def mock_token_manager():
+    """Mock the token_manager used by user.py routes."""
+    with patch('server.routes.user.token_manager') as mock_tm:
+        yield mock_tm
+
+
+@pytest.fixture
+def mock_check_idp():
+    """Mock _check_idp to pass through the default_value (the fallback User).
+
+    This isolates the test to just the User construction logic in saas_get_user,
+    without needing to set up Keycloak IDP token checks.
+    """
+    with patch('server.routes.user._check_idp') as mock_fn:
+        # Return the default_value argument that was passed to _check_idp
+        mock_fn.side_effect = lambda **kwargs: kwargs.get('default_value')
+        yield mock_fn
+
+
+@pytest.mark.asyncio
+async def test_fallback_user_includes_name_from_name_claim(
+    mock_token_manager, mock_check_idp
+):
+    """When Keycloak provides a 'name' claim, the fallback User should include it."""
+    from server.routes.user import saas_get_user
+
+    mock_token_manager.get_user_info = AsyncMock(
+        return_value={
+            'sub': '248289761001',
+            'name': 'Jane Doe',
+            'preferred_username': 'j.doe',
+            'email': 'jane@example.com',
+        }
+    )
+
+    result = await saas_get_user(
+        provider_tokens=None,
+        access_token=SecretStr('test-token'),
+        user_id='248289761001',
+    )
+
+    assert isinstance(result, User)
+    assert result.name == 'Jane Doe'
+    assert result.email == 'jane@example.com'
+
+
+@pytest.mark.asyncio
+async def test_fallback_user_combines_given_and_family_name(
+    mock_token_manager, mock_check_idp
+):
+    """When 'name' is absent, combine given_name + family_name."""
+    from server.routes.user import saas_get_user
+
+    mock_token_manager.get_user_info = AsyncMock(
+        return_value={
+            'sub': '248289761001',
+            'given_name': 'Jane',
+            'family_name': 'Doe',
+            'preferred_username': 'j.doe',
+            'email': 'jane@example.com',
+        }
+    )
+
+    result = await saas_get_user(
+        provider_tokens=None,
+        access_token=SecretStr('test-token'),
+        user_id='248289761001',
+    )
+
+    assert isinstance(result, User)
+    assert result.name == 'Jane Doe'
+
+
+@pytest.mark.asyncio
+async def test_fallback_user_name_is_none_when_no_name_claims(
+    mock_token_manager, mock_check_idp
+):
+    """When no name claims exist, name should be None."""
+    from server.routes.user import saas_get_user
+
+    mock_token_manager.get_user_info = AsyncMock(
+        return_value={
+            'sub': '248289761001',
+            'preferred_username': 'j.doe',
+            'email': 'jane@example.com',
+        }
+    )
+
+    result = await saas_get_user(
+        provider_tokens=None,
+        access_token=SecretStr('test-token'),
+        user_id='248289761001',
+    )
+
+    assert isinstance(result, User)
+    assert result.name is None
+
+
+@pytest.mark.asyncio
+async def test_fallback_user_includes_company_claim(mock_token_manager, mock_check_idp):
+    """When Keycloak provides a 'company' claim, include it in the User."""
+    from server.routes.user import saas_get_user
+
+    mock_token_manager.get_user_info = AsyncMock(
+        return_value={
+            'sub': '248289761001',
+            'name': 'Jane Doe',
+            'preferred_username': 'j.doe',
+            'email': 'jane@example.com',
+            'company': 'Acme Corp',
+        }
+    )
+
+    result = await saas_get_user(
+        provider_tokens=None,
+        access_token=SecretStr('test-token'),
+        user_id='248289761001',
+    )
+
+    assert isinstance(result, User)
+    assert result.company == 'Acme Corp'
+
+
+@pytest.mark.asyncio
+async def test_fallback_user_company_is_none_when_absent(
+    mock_token_manager, mock_check_idp
+):
+    """When 'company' is not in Keycloak claims, company should be None."""
+    from server.routes.user import saas_get_user
+
+    mock_token_manager.get_user_info = AsyncMock(
+        return_value={
+            'sub': '248289761001',
+            'name': 'Jane Doe',
+            'preferred_username': 'j.doe',
+            'email': 'jane@example.com',
+        }
+    )
+
+    result = await saas_get_user(
+        provider_tokens=None,
+        access_token=SecretStr('test-token'),
+        user_id='248289761001',
+    )
+
+    assert isinstance(result, User)
+    assert result.company is None

enterprise/tests/unit/test_user_store.py

@@ -1,16 +1,16 @@
 import uuid
+from contextlib import asynccontextmanager
 from unittest.mock import AsyncMock, MagicMock, patch
 
 import pytest
 from pydantic import SecretStr
-
-# Mock the database module before importing UserStore
-with patch('storage.database.engine'), patch('storage.database.a_engine'):
-    from storage.user import User
-    from storage.user_store import UserStore
-
 from sqlalchemy.orm import configure_mappers
 
+# Database connection is lazy (no module-level engines), so no patching needed
+from storage.org import Org
+from storage.user import User
+from storage.user_store import UserStore
+
 from openhands.storage.data_models.settings import Settings
 
 
@@ -162,3 +162,406 @@ def test_get_kwargs_from_settings():
     assert 'enable_sound_notifications' in kwargs
     # Should not include fields that don't exist in User model
     assert 'llm_api_key' not in kwargs
+
+
+# --- Tests for contact_name resolution in migrate_user() ---
+# migrate_user() should use resolve_display_name() to populate contact_name
+# from Keycloak name claims, falling back to username only when no real name
+# is available. This mirrors the create_user() fix and ensures migrated Org
+# records also store the user's actual display name.
+
+
+class _StopAfterOrgCreation(Exception):
+    """Halt migrate_user() after Org creation for contact_name inspection."""
+
+    pass
+
+
+@pytest.mark.asyncio
+async def test_migrate_user_contact_name_uses_name_claim():
+    """When user_info has a 'name' claim, migrate_user() should use it for contact_name."""
+    user_id = str(uuid.uuid4())
+    user_info = {
+        'username': 'jdoe',
+        'email': 'jdoe@example.com',
+        'name': 'John Doe',
+    }
+
+    mock_session = MagicMock()
+    mock_sm = MagicMock()
+    mock_sm.return_value.__enter__ = MagicMock(return_value=mock_session)
+    mock_sm.return_value.__exit__ = MagicMock(return_value=False)
+
+    mock_user_settings = MagicMock()
+    mock_user_settings.user_version = 1
+
+    with (
+        patch('storage.user_store.session_maker', mock_sm),
+        patch(
+            'storage.user_store.decrypt_legacy_model',
+            return_value={'keycloak_user_id': user_id},
+        ),
+        patch('storage.user_store.UserSettings'),
+        patch(
+            'storage.lite_llm_manager.LiteLlmManager.migrate_entries',
+            new_callable=AsyncMock,
+            side_effect=_StopAfterOrgCreation,
+        ),
+    ):
+        with pytest.raises(_StopAfterOrgCreation):
+            await UserStore.migrate_user(user_id, mock_user_settings, user_info)
+
+    org = mock_session.add.call_args_list[0][0][0]
+    assert isinstance(org, Org)
+    assert org.contact_name == 'John Doe'
+
+
+@pytest.mark.asyncio
+async def test_migrate_user_contact_name_uses_given_family_names():
+    """When only given_name and family_name are present, migrate_user() should combine them."""
+    user_id = str(uuid.uuid4())
+    user_info = {
+        'username': 'jsmith',
+        'email': 'jsmith@example.com',
+        'given_name': 'Jane',
+        'family_name': 'Smith',
+    }
+
+    mock_session = MagicMock()
+    mock_sm = MagicMock()
+    mock_sm.return_value.__enter__ = MagicMock(return_value=mock_session)
+    mock_sm.return_value.__exit__ = MagicMock(return_value=False)
+
+    mock_user_settings = MagicMock()
+    mock_user_settings.user_version = 1
+
+    with (
+        patch('storage.user_store.session_maker', mock_sm),
+        patch(
+            'storage.user_store.decrypt_legacy_model',
+            return_value={'keycloak_user_id': user_id},
+        ),
+        patch('storage.user_store.UserSettings'),
+        patch(
+            'storage.lite_llm_manager.LiteLlmManager.migrate_entries',
+            new_callable=AsyncMock,
+            side_effect=_StopAfterOrgCreation,
+        ),
+    ):
+        with pytest.raises(_StopAfterOrgCreation):
+            await UserStore.migrate_user(user_id, mock_user_settings, user_info)
+
+    org = mock_session.add.call_args_list[0][0][0]
+    assert isinstance(org, Org)
+    assert org.contact_name == 'Jane Smith'
+
+
+@pytest.mark.asyncio
+async def test_migrate_user_contact_name_falls_back_to_username():
+    """When no name claims exist, migrate_user() should fall back to username."""
+    user_id = str(uuid.uuid4())
+    user_info = {
+        'username': 'jdoe',
+        'email': 'jdoe@example.com',
+    }
+
+    mock_session = MagicMock()
+    mock_sm = MagicMock()
+    mock_sm.return_value.__enter__ = MagicMock(return_value=mock_session)
+    mock_sm.return_value.__exit__ = MagicMock(return_value=False)
+
+    mock_user_settings = MagicMock()
+    mock_user_settings.user_version = 1
+
+    with (
+        patch('storage.user_store.session_maker', mock_sm),
+        patch(
+            'storage.user_store.decrypt_legacy_model',
+            return_value={'keycloak_user_id': user_id},
+        ),
+        patch('storage.user_store.UserSettings'),
+        patch(
+            'storage.lite_llm_manager.LiteLlmManager.migrate_entries',
+            new_callable=AsyncMock,
+            side_effect=_StopAfterOrgCreation,
+        ),
+    ):
+        with pytest.raises(_StopAfterOrgCreation):
+            await UserStore.migrate_user(user_id, mock_user_settings, user_info)
+
+    org = mock_session.add.call_args_list[0][0][0]
+    assert isinstance(org, Org)
+    assert org.contact_name == 'jdoe'
+
+
+# --- Tests for contact_name resolution in create_user() ---
+# create_user() should use resolve_display_name() to populate contact_name
+# from Keycloak name claims, falling back to preferred_username only when
+# no real name is available. This ensures Org records store the user's
+# actual display name for use in UI and analytics.
+
+
+@pytest.mark.asyncio
+async def test_create_user_contact_name_uses_name_claim():
+    """When user_info has a 'name' claim, create_user() should use it for contact_name."""
+    user_id = str(uuid.uuid4())
+    user_info = {
+        'preferred_username': 'jdoe',
+        'email': 'jdoe@example.com',
+        'name': 'John Doe',
+    }
+
+    mock_session = MagicMock()
+    mock_sm = MagicMock()
+    mock_sm.return_value.__enter__ = MagicMock(return_value=mock_session)
+    mock_sm.return_value.__exit__ = MagicMock(return_value=False)
+
+    with (
+        patch('storage.user_store.session_maker', mock_sm),
+        patch.object(
+            UserStore,
+            'create_default_settings',
+            new_callable=AsyncMock,
+            return_value=None,
+        ),
+    ):
+        result = await UserStore.create_user(user_id, user_info)
+
+    assert result is None  # create_default_settings returned None
+    # The Org should have been added to the session with the real display name
+    org = mock_session.add.call_args_list[0][0][0]
+    assert isinstance(org, Org)
+    assert org.contact_name == 'John Doe'
+
+
+@pytest.mark.asyncio
+async def test_create_user_contact_name_uses_given_family_names():
+    """When only given_name and family_name are present, create_user() should combine them."""
+    user_id = str(uuid.uuid4())
+    user_info = {
+        'preferred_username': 'jsmith',
+        'email': 'jsmith@example.com',
+        'given_name': 'Jane',
+        'family_name': 'Smith',
+    }
+
+    mock_session = MagicMock()
+    mock_sm = MagicMock()
+    mock_sm.return_value.__enter__ = MagicMock(return_value=mock_session)
+    mock_sm.return_value.__exit__ = MagicMock(return_value=False)
+
+    with (
+        patch('storage.user_store.session_maker', mock_sm),
+        patch.object(
+            UserStore,
+            'create_default_settings',
+            new_callable=AsyncMock,
+            return_value=None,
+        ),
+    ):
+        result = await UserStore.create_user(user_id, user_info)
+
+    assert result is None
+    org = mock_session.add.call_args_list[0][0][0]
+    assert isinstance(org, Org)
+    assert org.contact_name == 'Jane Smith'
+
+
+@pytest.mark.asyncio
+async def test_create_user_contact_name_falls_back_to_username():
+    """When no name claims exist, create_user() should fall back to preferred_username."""
+    user_id = str(uuid.uuid4())
+    user_info = {
+        'preferred_username': 'jdoe',
+        'email': 'jdoe@example.com',
+    }
+
+    mock_session = MagicMock()
+    mock_sm = MagicMock()
+    mock_sm.return_value.__enter__ = MagicMock(return_value=mock_session)
+    mock_sm.return_value.__exit__ = MagicMock(return_value=False)
+
+    with (
+        patch('storage.user_store.session_maker', mock_sm),
+        patch.object(
+            UserStore,
+            'create_default_settings',
+            new_callable=AsyncMock,
+            return_value=None,
+        ),
+    ):
+        result = await UserStore.create_user(user_id, user_info)
+
+    assert result is None
+    org = mock_session.add.call_args_list[0][0][0]
+    assert isinstance(org, Org)
+    assert org.contact_name == 'jdoe'
+
+
+# --- Tests for backfill_contact_name on login ---
+# Existing users created before the resolve_display_name fix may have
+# username-style values in contact_name. The backfill updates these to
+# the user's real display name when they next log in, but preserves
+# custom values set via the PATCH endpoint.
+
+
+def _wrap_sync_as_async_session_maker(sync_sm):
+    """Wrap a sync session_maker so it can be used in place of a_session_maker."""
+
+    @asynccontextmanager
+    async def _async_sm():
+        session = sync_sm()
+        try:
+
+            class _AsyncWrapper:
+                async def execute(self, *args, **kwargs):
+                    return session.execute(*args, **kwargs)
+
+                async def commit(self):
+                    session.commit()
+
+            yield _AsyncWrapper()
+        finally:
+            session.close()
+
+    return _async_sm
+
+
+@pytest.mark.asyncio
+async def test_backfill_contact_name_updates_when_matches_preferred_username(
+    session_maker,
+):
+    """When contact_name matches preferred_username and a real name is available, update it."""
+    user_id = str(uuid.uuid4())
+    # Create org with username-style contact_name (as create_user used to store)
+    with session_maker() as session:
+        org = Org(
+            id=uuid.UUID(user_id),
+            name=f'user_{user_id}_org',
+            contact_name='jdoe',
+            contact_email='jdoe@example.com',
+        )
+        session.add(org)
+        session.commit()
+
+    user_info = {
+        'preferred_username': 'jdoe',
+        'name': 'John Doe',
+    }
+
+    with patch(
+        'storage.user_store.a_session_maker',
+        _wrap_sync_as_async_session_maker(session_maker),
+    ):
+        await UserStore.backfill_contact_name(user_id, user_info)
+
+    with session_maker() as session:
+        org = session.query(Org).filter(Org.id == uuid.UUID(user_id)).first()
+        assert org.contact_name == 'John Doe'
+
+
+@pytest.mark.asyncio
+async def test_backfill_contact_name_updates_when_matches_username(session_maker):
+    """When contact_name matches username (migrate_user legacy) and a real name is available, update it."""
+    user_id = str(uuid.uuid4())
+    # Create org with username-style contact_name (as migrate_user used to store)
+    with session_maker() as session:
+        org = Org(
+            id=uuid.UUID(user_id),
+            name=f'user_{user_id}_org',
+            contact_name='jdoe',
+            contact_email='jdoe@example.com',
+        )
+        session.add(org)
+        session.commit()
+
+    user_info = {
+        'username': 'jdoe',
+        'given_name': 'Jane',
+        'family_name': 'Doe',
+    }
+
+    with patch(
+        'storage.user_store.a_session_maker',
+        _wrap_sync_as_async_session_maker(session_maker),
+    ):
+        await UserStore.backfill_contact_name(user_id, user_info)
+
+    with session_maker() as session:
+        org = session.query(Org).filter(Org.id == uuid.UUID(user_id)).first()
+        assert org.contact_name == 'Jane Doe'
+
+
+@pytest.mark.asyncio
+async def test_backfill_contact_name_preserves_custom_value(session_maker):
+    """When contact_name differs from both username fields, do not overwrite it."""
+    user_id = str(uuid.uuid4())
+    # Org has a custom contact_name set via PATCH endpoint
+    with session_maker() as session:
+        org = Org(
+            id=uuid.UUID(user_id),
+            name=f'user_{user_id}_org',
+            contact_name='Custom Corp Name',
+            contact_email='jdoe@example.com',
+        )
+        session.add(org)
+        session.commit()
+
+    user_info = {
+        'preferred_username': 'jdoe',
+        'username': 'jdoe',
+        'name': 'John Doe',
+    }
+
+    with patch(
+        'storage.user_store.a_session_maker',
+        _wrap_sync_as_async_session_maker(session_maker),
+    ):
+        await UserStore.backfill_contact_name(user_id, user_info)
+
+    with session_maker() as session:
+        org = session.query(Org).filter(Org.id == uuid.UUID(user_id)).first()
+        assert org.contact_name == 'Custom Corp Name'
+
+
+def test_update_current_org_success(session_maker):
+    """
+    GIVEN: User exists in database
+    WHEN: update_current_org is called with new org_id
+    THEN: User's current_org_id is updated and user is returned
+    """
+    # Arrange
+    user_id = str(uuid.uuid4())
+    initial_org_id = uuid.uuid4()
+    new_org_id = uuid.uuid4()
+
+    with session_maker() as session:
+        user = User(id=uuid.UUID(user_id), current_org_id=initial_org_id)
+        session.add(user)
+        session.commit()
+
+    # Act
+    with patch('storage.user_store.session_maker', session_maker):
+        result = UserStore.update_current_org(user_id, new_org_id)
+
+    # Assert
+    assert result is not None
+    assert result.current_org_id == new_org_id
+
+
+def test_update_current_org_user_not_found(session_maker):
+    """
+    GIVEN: User does not exist in database
+    WHEN: update_current_org is called
+    THEN: None is returned
+    """
+    # Arrange
+    user_id = str(uuid.uuid4())
+    org_id = uuid.uuid4()
+
+    # Act
+    with patch('storage.user_store.session_maker', session_maker):
+        result = UserStore.update_current_org(user_id, org_id)
+
+    # Assert
+    assert result is None

enterprise/utils/identity.py

@@ -0,0 +1,25 @@
+"""Shared identity helpers for resolving user profile fields from Keycloak claims."""
+
+
+def resolve_display_name(user_info: dict) -> str | None:
+    """Resolve the best available display name from Keycloak user_info claims.
+
+    Fallback chain: name → given_name + family_name → None
+
+    Does NOT fall back to preferred_username/username — callers that need
+    a guaranteed non-None value should handle that separately. This keeps
+    the helper focused on real-name claims so that the /api/user/info route
+    can return name=None when no real name is available, while user_store
+    callers can append their own username fallback.
+    """
+    name = user_info.get('name', '')
+    if name and name.strip():
+        return name.strip()
+
+    given = user_info.get('given_name', '').strip()
+    family = user_info.get('family_name', '').strip()
+    combined = f'{given} {family}'.strip()
+    if combined:
+        return combined
+
+    return None

evaluation/README.md

@@ -1,152 +0,0 @@
-# Evaluation
-
-> [!WARNING]
-> **This directory is deprecated.** Our new benchmarks are located at [OpenHands/benchmarks](https://github.com/OpenHands/benchmarks).
->
-> If you have already implemented a benchmark in this directory and would like to contribute it, we are happy to have the contribution. However, if you are starting anew, please use the new location.
-
-This folder contains code and resources to run experiments and evaluations.
-
-## For Benchmark Users
-
-### Setup
-
-Before starting evaluation, follow the instructions [here](https://github.com/OpenHands/OpenHands/blob/main/Development.md) to setup your local development environment and LLM.
-
-Once you are done with setup, you can follow the benchmark-specific instructions in each subdirectory of the [evaluation directory](#supported-benchmarks).
-Generally these will involve running `run_infer.py` to perform inference with the agents.
-
-### Implementing and Evaluating an Agent
-
-To add an agent to OpenHands, you will need to implement it in the [agenthub directory](https://github.com/OpenHands/OpenHands/tree/main/openhands/agenthub). There is a README there with more information.
-
-To evaluate an agent, you can provide the agent's name to the `run_infer.py` program.
-
-### Evaluating Different LLMs
-
-OpenHands in development mode uses `config.toml` to keep track of most configuration.
-**IMPORTANT: For evaluation, only the LLM section in `config.toml` will be used. Other configurations, such as `save_trajectory_path`, are not applied during evaluation.**
-
-Here's an example configuration file you can use to define and use multiple LLMs:
-
-```toml
-[llm]
-# IMPORTANT: add your API key here, and set the model to the one you want to evaluate
-model = "gpt-4o-2024-05-13"
-api_key = "sk-XXX"
-
-[llm.eval_gpt4_1106_preview_llm]
-model = "gpt-4-1106-preview"
-api_key = "XXX"
-temperature = 0.0
-
-[llm.eval_some_openai_compatible_model_llm]
-model = "openai/MODEL_NAME"
-base_url = "https://OPENAI_COMPATIBLE_URL/v1"
-api_key = "XXX"
-temperature = 0.0
-```
-
-### Configuring Condensers for Evaluation
-
-For benchmarks that support condenser configuration (like SWE-Bench), you can define multiple condenser configurations in your `config.toml` file. A condenser is responsible for managing conversation history to maintain context while staying within token limits - you can learn more about how it works [here](https://www.openhands.dev/blog/openhands-context-condensensation-for-more-efficient-ai-agents):
-
-```toml
-# LLM-based summarizing condenser for evaluation
-[condenser.summarizer_for_eval]
-type = "llm"
-llm_config = "haiku"  # Reference to an LLM config to use for summarization
-keep_first = 2        # Number of initial events to always keep
-max_size = 100        # Maximum size of history before triggering summarization
-
-# Recent events condenser for evaluation
-[condenser.recent_for_eval]
-type = "recent"
-keep_first = 2        # Number of initial events to always keep
-max_events = 50       # Maximum number of events to keep in history
-```
-
-You can then specify which condenser configuration to use when running evaluation scripts, for example:
-
-```bash
-EVAL_CONDENSER=summarizer_for_eval \
-./evaluation/benchmarks/swe_bench/scripts/run_infer.sh llm.eval_gpt4_1106_preview HEAD CodeActAgent 500 100 1 princeton-nlp/SWE-bench_Verified test
-```
-
-The name is up to you, but should match a name defined in your `config.toml` file. The last argument in the command specifies the condenser configuration to use. In this case, `summarizer_for_eval` is used, which refers to the LLM-based summarizing condenser as defined above.
-
-If no condenser configuration is specified, the 'noop' condenser will be used by default, which keeps the full conversation history.
-
-For other configurations specific to evaluation, such as `save_trajectory_path`, these are typically set in the `get_config` function of the respective `run_infer.py` file for each benchmark.
-
-### Enabling LLM-Based Editor Tools
-
-The LLM-Based Editor tool (currently supported only for SWE-Bench) can be enabled by setting:
-```bash
-export ENABLE_LLM_EDITOR=true
-```
-
-You can set the config for the Editor LLM as:
-```toml
-[llm.draft_editor]
-base_url = "http://localhost:9002/v1"
-model = "hosted_vllm/lite_coder_qwen_editor_3B"
-api_key = ""
-temperature = 0.7
-max_input_tokens = 10500
-max_output_tokens = 10500
-```
-
-## Supported Benchmarks
-
-The OpenHands evaluation harness supports a wide variety of benchmarks across [software engineering](#software-engineering), [web browsing](#web-browsing), [miscellaneous assistance](#misc-assistance), and [real-world](#real-world) tasks.
-
-### Software Engineering
-
-- SWE-Bench: [`evaluation/benchmarks/swe_bench`](./benchmarks/swe_bench)
-- HumanEvalFix: [`evaluation/benchmarks/humanevalfix`](./benchmarks/humanevalfix)
-- BIRD: [`evaluation/benchmarks/bird`](./benchmarks/bird)
-- BioCoder: [`evaluation/benchmarks/biocoder`](./benchmarks/biocoder)
-- ML-Bench: [`evaluation/benchmarks/ml_bench`](./benchmarks/ml_bench)
-- APIBench: [`evaluation/benchmarks/gorilla`](./benchmarks/gorilla/)
-- ToolQA: [`evaluation/benchmarks/toolqa`](./benchmarks/toolqa/)
-- AiderBench: [`evaluation/benchmarks/aider_bench`](./benchmarks/aider_bench/)
-- Commit0: [`evaluation/benchmarks/commit0_bench`](./benchmarks/commit0_bench/)
-- DiscoveryBench: [`evaluation/benchmarks/discoverybench`](./benchmarks/discoverybench/)
-- TerminalBench: [`evaluation/benchmarks/terminal_bench`](./benchmarks/terminal_bench)
-
-### Web Browsing
-
-- WebArena: [`evaluation/benchmarks/webarena`](./benchmarks/webarena/)
-- MiniWob++: [`evaluation/benchmarks/miniwob`](./benchmarks/miniwob/)
-- Browsing Delegation: [`evaluation/benchmarks/browsing_delegation`](./benchmarks/browsing_delegation/)
-
-### Misc. Assistance
-
-- GAIA: [`evaluation/benchmarks/gaia`](./benchmarks/gaia)
-- GPQA: [`evaluation/benchmarks/gpqa`](./benchmarks/gpqa)
-- AgentBench: [`evaluation/benchmarks/agent_bench`](./benchmarks/agent_bench)
-- MINT: [`evaluation/benchmarks/mint`](./benchmarks/mint)
-- Entity deduction Arena (EDA): [`evaluation/benchmarks/EDA`](./benchmarks/EDA)
-- ProofWriter: [`evaluation/benchmarks/logic_reasoning`](./benchmarks/logic_reasoning)
-- ScienceAgentBench: [`evaluation/benchmarks/scienceagentbench`](./benchmarks/scienceagentbench)
-
-### Real World
-
-- TheAgentCompany: [`evaluation/benchmarks/the_agent_company`](./benchmarks/the_agent_company)
-
-## Result Visualization
-
-Check [this huggingface space](https://huggingface.co/spaces/OpenHands/evaluation) for visualization of existing experimental results.
-
-You can start your own fork of [our huggingface evaluation outputs](https://huggingface.co/spaces/OpenHands/evaluation) and submit a PR of your evaluation results to our hosted huggingface repo via PR following the guide [here](https://huggingface.co/docs/hub/en/repositories-pull-requests-discussions#pull-requests-and-discussions).
-
-## For Benchmark Developers
-
-To learn more about how to integrate your benchmark into OpenHands, check out [tutorial here](https://docs.openhands.dev/usage/how-to/evaluation-harness). Briefly,
-
-- Each subfolder contains a specific benchmark or experiment. For example, [`evaluation/benchmarks/swe_bench`](./benchmarks/swe_bench) should contain
-all the preprocessing/evaluation/analysis scripts.
-- Raw data and experimental records should not be stored within this repo.
-- For model outputs, they should be stored at [this huggingface space](https://huggingface.co/spaces/OpenHands/evaluation) for visualization.
-- Important data files of manageable size and analysis scripts (e.g., jupyter notebooks) can be directly uploaded to this repo.

evaluation/benchmarks/EDA/README.md

@@ -1,46 +0,0 @@
-# EDA Evaluation
-
-This folder contains evaluation harness for evaluating agents on the Entity-deduction-Arena Benchmark, from the paper [Probing the Multi-turn Planning Capabilities of LLMs via 20 Question Games](https://arxiv.org/abs/2310.01468), presented in ACL 2024 main conference.
-
-## Setup Environment and LLM Configuration
-
-Please follow instruction [here](../../README.md#setup) to setup your local development environment and LLM.
-
-## Start the evaluation
-
-```bash
-export OPENAI_API_KEY="sk-XXX"; # This is required for evaluation (to simulate another party of conversation)
-./evaluation/benchmarks/EDA/scripts/run_infer.sh [model_config] [git-version] [agent] [dataset] [eval_limit]
-```
-
-where `model_config` is mandatory, while `git-version`, `agent`, `dataset` and `eval_limit` are optional.
-
-- `model_config`, e.g. `eval_gpt4_1106_preview`, is the config group name for your
-LLM settings, as defined in your `config.toml`.
-
-- `git-version`, e.g. `HEAD`, is the git commit hash of the OpenHands version you would
-like to evaluate. It could also be a release tag like `0.6.2`.
-
-- `agent`, e.g. `CodeActAgent`, is the name of the agent for benchmarks, defaulting
-to `CodeActAgent`.
-
-- `dataset`: There are two tasks in this evaluation. Specify `dataset` to test on either `things` or `celebs` task.
-
-- `eval_limit`, e.g. `10`, limits the evaluation to the first `eval_limit` instances. By default it infers all instances.
-
-For example,
-
-```bash
-./evaluation/benchmarks/EDA/scripts/run_infer.sh eval_gpt4o_2024_05_13 0.6.2 CodeActAgent things
-```
-
-## Reference
-
-```bibtex
-@inproceedings{zhang2023entity,
-  title={Probing the Multi-turn Planning Capabilities of LLMs via 20 Question Games},
-  author={Zhang, Yizhe and Lu, Jiarui and Jaitly, Navdeep},
-  journal={ACL},
-  year={2024}
-}
-```

evaluation/benchmarks/EDA/game.py

@@ -1,203 +0,0 @@
-import logging
-import re
-
-import httpx
-import openai
-from openai import OpenAI
-from retry import retry
-
-LOGGER = logging.getLogger(__name__)
-
-
-class Q20Game:
-    def __init__(
-        self,
-        item: str,
-        answerer_model: str = 'gpt-3.5-turbo-0613',
-        guesser_model: str = 'gpt-3.5-turbo-0613',
-        num_turns: int = 20,
-        temperature: float = 0.8,
-        openai_api: bool = True,
-        openai_api_key: str | None = None,
-        guesser_kargs=None,
-    ) -> None:
-        if guesser_kargs is None:
-            guesser_kargs = {}
-        self.item = item
-        self.answerer_model = answerer_model
-        self.guesser_model = guesser_model
-        self.num_turns = num_turns
-        self.temperature = temperature
-        self.openai_api = openai_api
-        self.guesser_kargs = guesser_kargs
-        self.vicuna_prompt = "A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions."
-        self.first_user_utterance = (
-            'Your task is to ask a series of questions to deduce the entity '
-            "that I'm thinking of with as few queries as possible. "
-            "Only ask questions that can be answered by 'yes', 'no' or 'maybe'. "
-            'Do not ask for hint. Make your question brief with no linebreaker. '
-            'Now start asking a question.'
-        )
-        self.guesser_win = False
-        self.curr_turn = 0
-        if openai_api_key is not None:
-            openai.api_key = openai_api_key
-
-        if isinstance(answerer_model, str) and not answerer_model.startswith('gpt'):
-            self.user_api_base = 'http://0.0.0.0:8000/v1'
-        else:
-            self.user_api_base = 'https://api.openai.com/v1'
-
-        if isinstance(guesser_model, str) and not guesser_model.startswith('gpt'):
-            self.guesser_api_base = 'http://0.0.0.0:8000/v1'
-        else:
-            self.guesser_api_base = 'https://api.openai.com/v1'
-
-        self.guesser_messages = []
-
-    def preprocess_response(self, response):
-        response = re.sub(r'the entity you are thinking of', 'it', response)
-        response = re.sub(r"the entity you're thinking of", 'it', response)
-        response = re.sub(r" you're thinking of", '', response)
-        response = re.sub(r' you are thinking of', '', response)
-        return response
-
-    def judge_winner(self, response):
-        guesser_question = response.strip()
-
-        if self.curr_turn == self.num_turns - 1:
-            guesser_question += ' Is it right?'
-
-        self.guesser_messages.append({'role': 'assistant', 'content': guesser_question})
-        # ask for answer
-        usr_msg = self.answerer(guesser_question)
-
-        self.guesser_messages.append(
-            {'role': 'user', 'content': f'{usr_msg["content"].strip()}'}
-        )
-
-        if 'bingo' in usr_msg['content'].lower():
-            self.guesser_win = True
-            return True, ''
-
-        return False, usr_msg['content'].strip()
-
-    def generate_user_response(self, response):
-        response = self.preprocess_response(response)
-        # others
-        bingo, anwser_reply = self.judge_winner(response)
-        if bingo:
-            return 'You are bingo! Use the "finish" tool to finish the interaction.\n'
-        if self.curr_turn == self.num_turns - 2:
-            anwser_reply += " You must guess now, what's it?"
-        return anwser_reply
-
-    def reward(self):
-        if self.guesser_win:
-            n_turns = (len(self.guesser_messages) + 1) // 2
-            return 1 - max(n_turns - 5, 0) * 0.02
-        return 0
-
-    @retry(
-        (
-            openai.Timeout,
-            httpx.TimeoutException,
-            openai.RateLimitError,
-            openai.APIError,
-            openai.APIConnectionError,
-        ),
-        tries=5,
-        delay=0.5,
-        backoff=0.5,
-        max_delay=2,
-        logger=LOGGER,
-    )
-    def answerer(self, question):
-        openai.api_base = self.user_api_base
-        client = OpenAI(api_key=openai.api_key)
-        user_messages = [
-            {
-                'role': 'user',
-                'content': f'Based on your knowledge about {self.item}, '
-                f'respond to the following question or guess. '
-                f"Limit your respond to only 'Yes.', 'No.' or 'Maybe.', with no explanation or other words. "
-                f'Never say the answer {self.item} in your response. '
-                f"If the question is to solicit the answer, respond 'No.'.",
-            },
-            {
-                'role': 'user',
-                'content': f'For the entity {self.item}, {question} (Yes/No/Maybe)',
-            },
-        ]
-
-        response = client.chat.completions.create(
-            model=self.answerer_model,
-            messages=user_messages,
-            max_tokens=6,
-            n=1,
-            stop=None,
-            temperature=0.2,
-        )
-        if any(
-            [
-                re.search(rf'(?:^|\W){i.strip().lower()}(?:$|\W)', question.lower())
-                for i in self.item.lower().split('|')
-            ]
-        ):
-            response.choices[0].message.content = 'Bingo!'
-        return response.choices[0].message.to_dict()
-
-
-class Q20GameCelebrity(Q20Game):
-    def __init__(self, item: str, **kwargs) -> None:
-        super().__init__(item, **kwargs)
-        self.first_user_utterance = (
-            'Your task is to ask a series of questions to deduce the celebrity '
-            "that I'm thinking of with as few queries as possible. "
-            "Only ask factual questions that can be answered by 'Yes.', 'No.' or 'Dunno.'. Do not ask for hint. Make your question brief with no linebreaker. "
-            'Now start asking a question.'
-        )
-
-    @retry(
-        (
-            openai.Timeout,
-            httpx.TimeoutException,
-            openai.RateLimitError,
-            openai.APIError,
-            openai.APIConnectionError,
-        ),
-        tries=5,
-        delay=0.5,
-        backoff=0.5,
-        max_delay=2,
-        logger=LOGGER,
-    )
-    def answerer(self, question):
-        openai.api_base = self.user_api_base
-        client = OpenAI(api_key=openai.api_key)
-        user_messages = [
-            {
-                'role': 'system',
-                'content': f'Based on your knowledge about the celebrity: {self.item}, '
-                f'respond to the following question or guess. '
-                f"Limit your respond to only 'Yes.', 'No.' or 'Dunno.', with no explanation or other words. "
-                f"Never say the name {self.item} in your response. Do not say 'Dunno.' if it can be answered by 'Yes.' or 'No.' "
-                f"If the question is to solicit the answer, respond 'No.'.",
-            },
-            {
-                'role': 'user',
-                'content': f'For the celebrity {self.item}, {question}(Yes/No/Dunno)',
-            },
-        ]
-
-        response = client.chat.completions.create(
-            model=self.answerer_model,
-            messages=user_messages,
-            max_tokens=6,
-            n=1,
-            stop=None,
-            temperature=0.2,
-        )
-        if re.search(rf'(?:^|\W){self.item.lower()}(?:$|\W)', question.lower()):
-            response.choices[0].message.content = 'Bingo!'
-        return response.choices[0].message.to_dict()

evaluation/benchmarks/EDA/run_infer.py

@@ -1,234 +0,0 @@
-import asyncio
-import os
-
-import pandas as pd
-from datasets import load_dataset
-
-from evaluation.benchmarks.EDA.game import Q20Game, Q20GameCelebrity
-from evaluation.utils.shared import (
-    EvalMetadata,
-    EvalOutput,
-    compatibility_for_eval_history_pairs,
-    get_metrics,
-    get_openhands_config_for_eval,
-    make_metadata,
-    prepare_dataset,
-    reset_logger_for_multiprocessing,
-    run_evaluation,
-)
-from openhands.controller.state.state import State
-from openhands.core.config import (
-    OpenHandsConfig,
-    get_evaluation_parser,
-    get_llm_config_arg,
-)
-from openhands.core.logger import openhands_logger as logger
-from openhands.core.main import create_runtime, run_controller
-from openhands.events.action import MessageAction
-from openhands.utils.async_utils import call_async_from_sync
-
-game = None
-
-
-def codeact_user_response_eda(state: State) -> str:
-    global game
-    model_guess = ''
-
-    # retrieve the latest model message from history
-    if state.history:
-        last_agent_message = state.get_last_agent_message()
-        model_guess = last_agent_message.content if last_agent_message else ''
-
-    assert game is not None, 'Game is not initialized.'
-    msg = game.generate_user_response(model_guess)
-    game.curr_turn += 1
-    logger.info(f'Model guess: {model_guess}')
-    logger.info(f'Answer response: {msg}')
-    if 'bingo!' in msg.lower():
-        return '/exit'
-    return msg
-
-
-AGENT_CLS_TO_FAKE_USER_RESPONSE_FN = {
-    'CodeActAgent': codeact_user_response_eda,
-}
-
-AGENT_CLS_TO_INST_SUFFIX = {
-    'CodeActAgent': 'When you think you have solved the question, please first send your answer to user through message and then exit.\n'
-}
-
-
-def get_config(
-    metadata: EvalMetadata,
-) -> OpenHandsConfig:
-    # Create config with EDA-specific container image
-    config = get_openhands_config_for_eval(
-        metadata=metadata,
-        runtime='docker',
-    )
-
-    # Override the container image for EDA
-    config.sandbox.base_container_image = 'python:3.12-bookworm'
-
-    config.set_llm_config(metadata.llm_config)
-    agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.enable_prompt_extensions = False
-    return config
-
-
-def process_instance(
-    instance: pd.Series,
-    metadata: EvalMetadata,
-    reset_logger: bool = True,
-) -> EvalOutput:
-    config = get_config(metadata)
-    instance_id = instance['text'].strip()
-
-    # Setup the logger properly, so you can run multi-processing to parallelize the evaluation
-    if reset_logger:
-        log_dir = os.path.join(metadata.eval_output_dir, 'infer_logs')
-        reset_logger_for_multiprocessing(logger, instance_id, log_dir)
-    else:
-        logger.info(f'Starting evaluation for instance {instance_id}.')
-
-    # Prepare instruction
-    _game_class = {'eda-things': Q20Game, 'eda-celebs': Q20GameCelebrity}
-
-    guesser_kargs = {
-        'max_new_tokens': 64,
-        'temperature': 0.8,
-        'repetition_penalty': 1.0,
-        'do_sample': True,
-    }  # no penalty
-
-    # Use codeactagent as guesser_model
-    global game
-    assert metadata.dataset is not None
-    assert metadata.details is not None
-    game = _game_class[metadata.dataset](
-        item=instance['text'].strip(),
-        answerer_model=metadata.details['answerer_model'],
-        guesser_model=None,
-        num_turns=metadata.max_iterations,
-        openai_api_key=metadata.details['openai_api_key'],
-        guesser_kargs=guesser_kargs,
-    )
-
-    instruction = f'{game.first_user_utterance}'
-    logger.info(f'Instruction: {instruction}')
-    instruction += AGENT_CLS_TO_INST_SUFFIX[metadata.agent_class]
-
-    # Here's how you can run the agent (similar to the `main` function) and get the final task state
-    runtime = create_runtime(config)
-    call_async_from_sync(runtime.connect)
-
-    state: State | None = asyncio.run(
-        run_controller(
-            config=config,
-            initial_user_action=MessageAction(content=instruction),
-            runtime=runtime,
-            fake_user_response_fn=AGENT_CLS_TO_FAKE_USER_RESPONSE_FN[
-                metadata.agent_class
-            ],
-        )
-    )
-    # ======= Attempt to evaluate the agent's edits =======
-    # If you are working on simpler benchmark that only evaluates the final model output (e.g., in a MessageAction)
-    # You can simply get the LAST `MessageAction` from the returned `state.history` and parse it for evaluation.
-
-    if state is None:
-        raise ValueError('State should not be None.')
-
-    last_agent_message = state.get_last_agent_message()
-    final_message = last_agent_message.content if last_agent_message else ''
-
-    logger.info(f'Final message: {final_message} | Ground truth: {instance["text"]}')
-    test_result = game.reward()
-    metrics = get_metrics(state)
-
-    # history is now available as a stream of events, rather than list of pairs of (Action, Observation)
-    # for compatibility with the existing output format, we can remake the pairs here
-    # remove when it becomes unnecessary
-    histories = compatibility_for_eval_history_pairs(state.history)
-
-    # Save the output
-    output = EvalOutput(
-        instance_id=instance_id,
-        instance=instance.to_dict(),
-        instruction=instruction,
-        metadata=metadata,
-        history=histories,
-        metrics=metrics,
-        error=state.last_error if state and state.last_error else None,
-        test_result={
-            'success': test_result,
-            'final_message': final_message,
-            'ground_truth': instance['text'],
-        },
-    )
-    return output
-
-
-if __name__ == '__main__':
-    parser = get_evaluation_parser()
-    parser.add_argument(
-        '--answerer_model', '-a', default='gpt-3.5-turbo', help='answerer model'
-    )
-    parser.add_argument(
-        '--dataset',
-        default='things',
-        choices=['things', 'celebs'],
-        type=str,
-        help='dataset to be used',
-    )
-    parser.add_argument(
-        '--OPENAI_API_KEY', type=str, required=True, help='Your OpenAI API key'
-    )
-    parser.add_argument(
-        '--data-split',
-        default='test',
-        type=str,
-        help='data split, eg, test',
-    )
-    args, _ = parser.parse_known_args()
-
-    eda_dataset = load_dataset(
-        'yizheapple/entity-deduction-arena', name=args.dataset, split=args.data_split
-    )
-    eda_dataset.rename(columns={'text': 'instance_id'}, inplace=True)
-
-    llm_config = None
-    if args.llm_config:
-        llm_config = get_llm_config_arg(args.llm_config)
-        # modify_params must be False for evaluation purpose, for reproducibility and accurancy of results
-        llm_config.modify_params = False
-
-    if llm_config is None:
-        raise ValueError(f'Could not find LLM config: --llm_config {args.llm_config}')
-
-    metadata = make_metadata(
-        llm_config,
-        f'eda-{args.dataset}',
-        args.agent_cls,
-        args.max_iterations,
-        args.eval_note,
-        args.eval_output_dir,
-        data_split=args.data_split,
-        details={
-            'answerer_model': str(args.answerer_model),
-            'openai_api_key': str(args.OPENAI_API_KEY),
-        },
-    )
-
-    output_file = os.path.join(metadata.eval_output_dir, 'output.jsonl')
-    prepared_dataset = prepare_dataset(
-        eda_dataset.to_pandas(), output_file, args.eval_n_limit
-    )
-
-    run_evaluation(
-        prepared_dataset,
-        metadata,
-        output_file,
-        args.eval_num_workers,
-        process_instance,
-    )

evaluation/benchmarks/EDA/scripts/run_infer.sh

@@ -1,60 +0,0 @@
-#!/usr/bin/env bash
-set -eo pipefail
-
-source "evaluation/utils/version_control.sh"
-
-MODEL_CONFIG=$1
-COMMIT_HASH=$2
-AGENT=$3
-DATASET=$4
-EVAL_LIMIT=$5
-NUM_WORKERS=$6
-
-if [ -z "$NUM_WORKERS" ]; then
-  NUM_WORKERS=1
-  echo "Number of workers not specified, use default $NUM_WORKERS"
-fi
-checkout_eval_branch
-
-if [ -z "$AGENT" ]; then
-  echo "Agent not specified, use default CodeActAgent"
-  AGENT="CodeActAgent"
-fi
-
-get_openhands_version
-
-if [ -z "$DATASET" ]; then
-  echo "Dataset not specified, use default 'things'"
-  DATASET="things"
-fi
-
-# check if OPENAI_API_KEY is set
-if [ -z "$OPENAI_API_KEY" ]; then
-  echo "OPENAI_API_KEY is not set, please set it to run the script"
-  exit 1
-fi
-
-
-echo "AGENT: $AGENT"
-echo "OPENHANDS_VERSION: $OPENHANDS_VERSION"
-echo "MODEL_CONFIG: $MODEL_CONFIG"
-echo "DATASET: $DATASET"
-
-COMMAND="poetry run python evaluation/benchmarks/EDA/run_infer.py \
-  --agent-cls $AGENT \
-  --llm-config $MODEL_CONFIG \
-  --dataset $DATASET \
-  --data-split test \
-  --max-iterations 20 \
-  --OPENAI_API_KEY $OPENAI_API_KEY \
-  --eval-num-workers $NUM_WORKERS \
-  --eval-note ${OPENHANDS_VERSION}_${DATASET}"
-
-if [ -n "$EVAL_LIMIT" ]; then
-  echo "EVAL_LIMIT: $EVAL_LIMIT"
-  COMMAND="$COMMAND --eval-n-limit $EVAL_LIMIT"
-fi
-
-# Run the command
-echo $COMMAND
-eval $COMMAND

evaluation/benchmarks/agent_bench/README.md

@@ -1,56 +0,0 @@
-# AgentBench Evaluation
-
-This folder contains evaluation harness for evaluating agents on the [AgentBench: Evaluating LLMs as Agents](https://arxiv.org/abs/2308.03688). We currently only support running on the `osbench` subset.
-
-## Setup Environment and LLM Configuration
-
-Please follow instruction [here](../../README.md#setup) to setup your local development environment and LLM.
-
-## Start the evaluation
-
-```bash
-./evaluation/benchmarks/agent_bench/scripts/run_infer.sh [model_config] [git-version] [agent] [eval_limit]
-```
-
-- `model_config`, e.g. `eval_gpt4_1106_preview`, is the config group name for your
-LLM settings, as defined in your `config.toml`.
-- `git-version`, e.g. `HEAD`, is the git commit hash of the OpenHands version you would
-like to evaluate. It could also be a release tag like `0.6.2`.
-- `agent`, e.g. `CodeActAgent`, is the name of the agent for benchmarks, defaulting
-to `CodeActAgent`.
-- `eval_limit`, e.g. `10`, limits the evaluation to the first `eval_limit` instances. By
-default, the script evaluates the entire SWE-bench_Lite test set (300 issues). Note:
-in order to use `eval_limit`, you must also set `agent`.
-
-
-Following is the basic command to start the evaluation.
-
-You can update the arguments in the script `evaluation/benchmarks/agent_bench/scripts/run_infer.sh`, such as `--max-iterations`, `--eval-num-workers` and so on.
-
-- `--agent-cls`, the agent to use. For example, `CodeActAgent`.
-- `--llm-config`: the LLM configuration to use. For example, `eval_gpt4_1106_preview`.
-- `--max-iterations`: the number of iterations to run the evaluation. For example, `30`.
-- `--eval-num-workers`: the number of workers to use for evaluation. For example, `5`.
-- `--eval-n-limit`: the number of examples to evaluate. For example, `100`.
-
-```bash
-./evaluation/benchmarks/agent_bench/scripts/run_infer.sh eval_gpt35_turbo HEAD CodeActAgent 1
-```
-
-## Run with Remote Runtime (experimental)
-
-You can run the evaluation using a remote runtime instead of a local Docker container. This is useful when you want to run the evaluation in a cloud environment or when you don't have Docker installed locally.
-
-To use the remote runtime, set the following environment variables:
-
-```bash
-# Required environment variables
-export ALLHANDS_API_KEY="your-api-key"  # Contact the team to get an API key
-export RUNTIME=remote
-export SANDBOX_REMOTE_RUNTIME_API_URL="https://runtime.eval.all-hands.dev"
-
-# Run the evaluation
-./evaluation/benchmarks/agent_bench/scripts/run_infer.sh llm.eval_gpt4_1106_preview HEAD CodeActAgent 1
-```
-
-The remote runtime will build a container image and run the evaluation in a cloud environment. The results will be saved locally in the same way as when running with a local runtime.

evaluation/benchmarks/agent_bench/helper.py

@@ -1,77 +0,0 @@
-import os
-import re
-from functools import partial
-
-from evaluation.utils.shared import codeact_user_response
-from openhands.events.action import CmdRunAction, MessageAction
-
-
-def try_parse_answer(act) -> str | None:
-    raw_ans = ''
-    if isinstance(act, MessageAction) and act.source == 'agent':
-        raw_ans = act.content
-    elif isinstance(act, CmdRunAction) and act.source == 'agent':
-        raw_ans = act.thought
-    else:
-        return None
-    agent_answer = re.findall(r'<solution>(.*?)</solution>', raw_ans, re.DOTALL)
-    if not agent_answer:
-        return None
-    return agent_answer[0].strip()
-
-
-FAKE_RESPONSES = {
-    'CodeActAgent': partial(
-        codeact_user_response, encapsulate_solution=True, try_parse=try_parse_answer
-    ),
-}
-
-INST_SUFFIXES: dict[str, str] = {
-    'CodeActAgent': (
-        'When you think you have solved the question, '
-        'please first send your answer to user through message and then exit.\n'
-    )
-}
-
-
-def analysis_size(size_str):
-    size_str = size_str.strip()
-    avails = {
-        'B': 1,
-        'Byte': 1,
-        'K': 1024,
-        'KB': 1024,
-        'M': 1024 * 1024,
-        'MB': 1024 * 1024,
-        'G': 1024 * 1024 * 1024,
-        'GB': 1024 * 1024 * 1024,
-        'T': 1024 * 1024 * 1024 * 1024,
-        'TB': 1024 * 1024 * 1024 * 1024,
-        'P': 1024 * 1024 * 1024 * 1024 * 1024,
-        'PB': 1024 * 1024 * 1024 * 1024 * 1024,
-    }
-    for size_unit in avails:
-        if size_str.endswith(size_unit):
-            return int(size_str[: -len(size_unit)]) * avails[size_unit]
-    return int(size_str)
-
-
-def compare_results(check_method: str, model_answer: str, final_ans: str) -> bool:
-    try:
-        match check_method:
-            case 'check/integer-match.py':
-                return int(model_answer) == int(final_ans)
-            case 'check/size-match.py':
-                return analysis_size(model_answer) == analysis_size(final_ans)
-        return (
-            model_answer.replace('\r\n', '\n').replace('\r', '\n').strip()
-            == final_ans.replace('\r\n', '\n').replace('\r', '\n').strip()
-        )
-    except Exception:
-        return False
-
-
-def create_sh_file(filename: str, cmds: str) -> None:
-    with open(filename, 'w', encoding='utf-8') as file:
-        file.write(cmds.replace('\r\n', '\n'))
-    os.chmod(filename, 0o755)

evaluation/benchmarks/agent_bench/run_infer.py

@@ -1,318 +0,0 @@
-import asyncio
-import os
-import re
-import tempfile
-from typing import Any
-
-import pandas as pd
-from datasets import load_dataset
-
-from evaluation.benchmarks.agent_bench.helper import (
-    FAKE_RESPONSES,
-    INST_SUFFIXES,
-    compare_results,
-    create_sh_file,
-)
-from evaluation.utils.shared import (
-    EvalMetadata,
-    EvalOutput,
-    compatibility_for_eval_history_pairs,
-    get_metrics,
-    get_openhands_config_for_eval,
-    make_metadata,
-    prepare_dataset,
-    reset_logger_for_multiprocessing,
-    run_evaluation,
-)
-from openhands.controller.state.state import State
-from openhands.core.config import (
-    OpenHandsConfig,
-    get_llm_config_arg,
-    parse_arguments,
-)
-from openhands.core.logger import openhands_logger as logger
-from openhands.core.main import create_runtime, run_controller
-from openhands.events.action import AgentFinishAction, CmdRunAction, MessageAction
-from openhands.events.observation import CmdOutputObservation
-from openhands.runtime.base import Runtime
-from openhands.utils.async_utils import call_async_from_sync
-
-
-def get_config(
-    metadata: EvalMetadata,
-) -> OpenHandsConfig:
-    # Create config with agent_bench-specific container image
-    config = get_openhands_config_for_eval(metadata=metadata)
-
-    # Override the container image for agent_bench
-    config.sandbox.base_container_image = 'python:3.12-slim'
-
-    config.set_llm_config(metadata.llm_config)
-    agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.enable_prompt_extensions = False
-    return config
-
-
-def initialize_runtime(
-    runtime: Runtime,
-    instance: pd.Series,  # this argument is not required
-):
-    """Initialize the runtime for the agent.
-
-    This function is called before the runtime is used to run the agent.
-    """
-    logger.info(f'{"-" * 50} BEGIN Runtime Initialization Fn {"-" * 50}')
-    obs: CmdOutputObservation
-
-    # Set instance id
-    action = CmdRunAction(command='mkdir -p /workspace')
-    logger.info(action, extra={'msg_type': 'ACTION'})
-    obs = runtime.run_action(action)
-    assert obs.exit_code == 0
-
-    action = CmdRunAction(command='cd /workspace')
-    logger.info(action, extra={'msg_type': 'ACTION'})
-    obs = runtime.run_action(action)
-    assert obs.exit_code == 0
-
-    init_cmd = instance.init
-    if init_cmd is not None:
-        script_name = f'{instance.instance_id}_init.sh'
-
-        with tempfile.TemporaryDirectory() as tmpdir:
-            host_script_path = os.path.join(tmpdir, script_name)
-            create_sh_file(host_script_path, init_cmd)
-            runtime.copy_to(
-                host_script_path,
-                '/workspace',
-            )
-
-        logger.info(f'Running init script: {script_name}')
-        action = CmdRunAction(command=f'chmod +x ./{script_name} && ./{script_name}')
-        logger.info(action, extra={'msg_type': 'ACTION'})
-        obs = runtime.run_action(action)
-        logger.info(obs, extra={'msg_type': 'OBSERVATION'})
-        assert obs.exit_code == 0
-
-    logger.info(f'{"-" * 50} END Runtime Initialization Fn {"-" * 50}')
-
-
-def complete_runtime(
-    runtime: Runtime,
-    instance: pd.Series,  # this argument is not required, but it is used to get the workspace_dir_name
-) -> dict[str, Any]:
-    """Complete the runtime for the agent.
-
-    This function is called before the runtime is used to run the agent.
-    If you need to do something in the sandbox to get the correctness metric after
-    the agent has run, modify this function.
-    """
-    logger.info(f'{"-" * 50} BEGIN Runtime Completion Fn {"-" * 50}')
-    obs: CmdOutputObservation
-
-    agent_answer = None
-    get_agent_result_cmd = instance.get_agent_result
-    if get_agent_result_cmd is not None:
-        script_name = 'get_agent_result.sh'
-
-        with tempfile.TemporaryDirectory() as tmpdir:
-            host_script_path = os.path.join(tmpdir, script_name)
-            create_sh_file(host_script_path, get_agent_result_cmd)
-            runtime.copy_to(
-                host_script_path,
-                '/workspace',
-            )
-            logger.info(f'Running get agent result cmd: {script_name}')
-
-        action = CmdRunAction(
-            command=f'chmod +x ./{script_name} && ./{script_name}',
-        )
-        logger.info(action, extra={'msg_type': 'ACTION'})
-        obs = runtime.run_action(action)
-        logger.info(obs, extra={'msg_type': 'OBSERVATION'})
-        assert obs.exit_code == 0
-        agent_answer = obs.content
-    # IF the agent answer is not found, retrieve it from the history
-    # We wait until the controller finishes
-
-    final_ans = None
-    if instance.ground_truth is not None:
-        final_ans = instance.ground_truth
-    else:
-        get_ground_truth_cmd = instance.get_ground_truth
-        if get_ground_truth_cmd is not None:
-            script_name = 'get_ground_truth.sh'
-            with tempfile.TemporaryDirectory() as tmpdir:
-                host_script_path = os.path.join(tmpdir, script_name)
-                create_sh_file(host_script_path, get_ground_truth_cmd)
-                runtime.copy_to(
-                    host_script_path,
-                    '/workspace',
-                )
-            logger.info(f'Running get ground truth cmd: {script_name}')
-
-            action = CmdRunAction(
-                command=f'chmod +x ./{script_name} && ./{script_name}'
-            )
-            logger.info(action, extra={'msg_type': 'ACTION'})
-            obs = runtime.run_action(action)
-            logger.info(obs, extra={'msg_type': 'OBSERVATION'})
-            final_ans = obs.content
-
-    logger.info(f'{"-" * 50} END Runtime Completion Fn {"-" * 50}')
-    return {
-        'final_ans': final_ans,
-        'agent_answer': agent_answer,
-    }
-
-
-def process_instance(
-    instance: pd.Series,
-    metadata: EvalMetadata,
-    reset_logger: bool = True,
-) -> EvalOutput:
-    config = get_config(metadata)
-
-    # Setup the logger properly, so you can run multi-processing to parallelize the evaluation
-    if reset_logger:
-        log_dir = os.path.join(metadata.eval_output_dir, 'infer_logs')
-        reset_logger_for_multiprocessing(logger, instance.instance_id, log_dir)
-    else:
-        logger.info(f'Starting evaluation for instance {instance.instance_id}.')
-
-    # =============================================
-    # build instruction
-    # =============================================
-
-    # Prepare instruction
-    instruction = (
-        f'Please fix the following issue.\n'
-        'IMPORTANT: You should ONLY interact with the environment provided to you AND NEVER ASK FOR HUMAN HELP.\n'
-        'Please encapsulate your final answer (answer ONLY) within <solution> and </solution>.\n'
-        'For example: The answer to the question is <solution> 42 </solution>.\n'
-        '# Problem \n'
-        f'{instance.description}\n\n'
-    )
-    instruction += (
-        'IMPORTANT: You should ONLY interact with the environment provided '
-        'to you AND NEVER ASK FOR HUMAN HELP.\n'
-    )
-    # NOTE: You can actually set slightly different instruction for different agents
-    instruction += INST_SUFFIXES[metadata.agent_class]
-
-    # =============================================
-    # create sandbox and run the agent
-    # =============================================
-
-    runtime: Runtime = create_runtime(config)
-    call_async_from_sync(runtime.connect)
-
-    initialize_runtime(runtime, instance=instance)
-
-    # Here's how you can run the agent (similar to the `main` function) and get the final task state
-    state: State | None = asyncio.run(
-        run_controller(
-            config=config,
-            initial_user_action=MessageAction(content=instruction),
-            runtime=runtime,
-            fake_user_response_fn=FAKE_RESPONSES[metadata.agent_class],
-        )
-    )
-    if state is None:
-        raise ValueError('State should not be None.')
-
-    # =============================================
-    # result evaluation
-    # =============================================
-
-    return_val = complete_runtime(runtime, instance)
-    agent_answer = return_val['agent_answer']
-    final_ans = return_val['final_ans']
-
-    # If the agent answer is not found, retrieve it from the history
-    if agent_answer is None:
-        agent_answer = ''
-        logger.info('Retrieving agent answer from history.')
-        raw_ans = ''
-
-        # retrieve the last agent message or thought
-        for event in reversed(state.history):
-            if event.source == 'agent':
-                if isinstance(event, AgentFinishAction):
-                    raw_ans = event.thought
-                    break
-                elif isinstance(event, MessageAction):
-                    raw_ans = event.content
-                    break
-                elif isinstance(event, CmdRunAction):
-                    raw_ans = event.thought
-                    break
-
-        # parse the answer for a solution tag
-        agent_answer = re.findall(r'<solution>(.*?)</solution>', raw_ans, re.DOTALL)
-        if len(agent_answer) == 0:
-            logger.warning(f'Failed to parse model answer: {raw_ans}')
-            agent_answer = raw_ans
-        else:
-            agent_answer = agent_answer[0]
-
-    comparison_method = instance.comparison_method
-    logger.info(
-        f'Final message: {agent_answer} | Ground truth: {final_ans} | Comparison method: {comparison_method}'
-    )
-    test_result = compare_results(comparison_method, agent_answer, final_ans)
-
-    # history is now available as a stream of events, rather than list of pairs of (Action, Observation)
-    # for compatibility with the existing output format, we can remake the pairs here
-    # remove when it becomes unnecessary
-    histories = compatibility_for_eval_history_pairs(state.history)
-
-    metrics = get_metrics(state)
-
-    # Save the output
-    output = EvalOutput(
-        instance_id=instance.instance_id,
-        instance=instance.to_dict(),
-        instruction=instruction,
-        metadata=metadata,
-        history=histories,
-        metrics=metrics,
-        error=state.last_error if state and state.last_error else None,
-        test_result={
-            'agent_answer': agent_answer,
-            'final_answer': final_ans,
-            'check_method': comparison_method,
-            'result': test_result,
-        },
-    )
-    return output
-
-
-if __name__ == '__main__':
-    args = parse_arguments()
-    dataset = load_dataset('iFurySt/AgentBench')
-    agent_bench_tests = dataset['osbench'].to_pandas()
-
-    llm_config = None
-    if args.llm_config:
-        llm_config = get_llm_config_arg(args.llm_config)
-        # modify_params must be False for evaluation purpose, for reproducibility and accurancy of results
-        llm_config.modify_params = False
-
-    if llm_config is None:
-        raise ValueError(f'Could not find LLM config: --llm_config {args.llm_config}')
-
-    metadata = make_metadata(
-        llm_config,
-        'AgentBench-OS',
-        args.agent_cls,
-        args.max_iterations,
-        args.eval_note,
-        args.eval_output_dir,
-    )
-    output_file = os.path.join(metadata.eval_output_dir, 'output.jsonl')
-    instances = prepare_dataset(agent_bench_tests, output_file, args.eval_n_limit)
-
-    run_evaluation(
-        instances, metadata, output_file, args.eval_num_workers, process_instance
-    )

evaluation/benchmarks/agent_bench/scripts/run_infer.sh

@@ -1,42 +0,0 @@
-#!/usr/bin/env bash
-set -eo pipefail
-
-source "evaluation/utils/version_control.sh"
-
-MODEL_CONFIG=$1
-COMMIT_HASH=$2
-AGENT=$3
-EVAL_LIMIT=$4
-NUM_WORKERS=$5
-
-if [ -z "$NUM_WORKERS" ]; then
-  NUM_WORKERS=1
-  echo "Number of workers not specified, use default $NUM_WORKERS"
-fi
-checkout_eval_branch
-
-if [ -z "$AGENT" ]; then
-  echo "Agent not specified, use default CodeActAgent"
-  AGENT="CodeActAgent"
-fi
-
-get_openhands_version
-
-echo "AGENT: $AGENT"
-echo "OPENHANDS_VERSION: $OPENHANDS_VERSION"
-echo "MODEL_CONFIG: $MODEL_CONFIG"
-
-COMMAND="export PYTHONPATH=evaluation/benchmarks/agent_bench:\$PYTHONPATH && poetry run python evaluation/benchmarks/agent_bench/run_infer.py \
-  --agent-cls $AGENT \
-  --llm-config $MODEL_CONFIG \
-  --max-iterations 30 \
-  --eval-num-workers $NUM_WORKERS \
-  --eval-note $OPENHANDS_VERSION"
-
-if [ -n "$EVAL_LIMIT" ]; then
-  echo "EVAL_LIMIT: $EVAL_LIMIT"
-  COMMAND="$COMMAND --eval-n-limit $EVAL_LIMIT"
-fi
-
-# Run the command
-eval $COMMAND

evaluation/benchmarks/agent_bench/scripts/summarise_results.py

@@ -1,37 +0,0 @@
-import json
-import sys
-
-
-def extract_test_results(res_file_path: str) -> tuple[list[str], list[str]]:
-    passed = []
-    failed = []
-    with open(res_file_path, 'r') as file:
-        for line in file:
-            data = json.loads(line.strip())
-            instance_id = data['instance_id']
-            resolved = False
-            if 'test_result' in data and 'result' in data['test_result']:
-                resolved = data['test_result']['result']
-            if resolved:
-                passed.append(instance_id)
-            else:
-                failed.append(instance_id)
-    return passed, failed
-
-
-if __name__ == '__main__':
-    if len(sys.argv) != 2:
-        print(
-            'Usage: poetry run python summarise_results.py <path_to_output_jsonl_file>'
-        )
-        sys.exit(1)
-    json_file_path = sys.argv[1]
-    passed_tests, failed_tests = extract_test_results(json_file_path)
-    succ_rate = len(passed_tests) / (len(passed_tests) + len(failed_tests))
-    print(
-        f'\nPassed {len(passed_tests)} tests, failed {len(failed_tests)} tests, resolve rate = {succ_rate}'
-    )
-    print('PASSED TESTS:')
-    print(passed_tests)
-    print('FAILED TESTS:')
-    print(failed_tests)

evaluation/benchmarks/aider_bench/README.md

@@ -1,96 +0,0 @@
-# AiderBench Evaluation
-
-This folder contains evaluation harness for evaluating agents on the
-[Aider Editing Benchmark](https://github.com/paul-gauthier/aider/blob/main/benchmark/README.md).
-This will allow us to develop better editing approach without running the full
-SWE-bench. The benchmark uses the
-[RajMaheshwari/Exercism-Python](https://huggingface.co/datasets/RajMaheshwari/Exercism-Python)
-Hugging Face dataset based on the
-[Exercism python coding exercises](https://github.com/exercism/python).
-
-## Setup Environment and LLM Configuration
-
-Please follow instruction [here](../../README.md#setup) to setup your local
-development environment and LLM.
-
-## Start the evaluation
-
-```bash
-./evaluation/benchmarks/aider_bench/scripts/run_infer.sh [model_config] [git-version] [agent] [eval_limit] [eval-num-workers] [eval_ids]
-```
-
-- `model_config`, e.g. `eval_gpt4_1106_preview`, is the config group name for
-    your LLM settings, as defined in your `config.toml`.
-- `git-version`, e.g. `HEAD`, is the git commit hash of the OpenHands version
-    you would like to evaluate. It could also be a release tag like `0.9.0`.
-- `agent`, e.g. `CodeActAgent`, is the name of the agent for benchmarks,
-    defaulting to `CodeActAgent`.
-- `eval_limit`, e.g. `10`, limits the evaluation to the first `eval_limit`
-    instances. By default, the script evaluates the entire Exercism test set
-    (133 issues). Note: in order to use `eval_limit`, you must also set `agent`.
-- `eval-num-workers`: the number of workers to use for evaluation. Default: `1`.
-- `eval_ids`, e.g. `"1,3,10"`, limits the evaluation to instances with the
-    given IDs (comma separated).
-
-There are also following optional environment variables you can set:
-
-```bash
-export USE_UNIT_TESTS=true # if you want to allow the Agent to verify correctness using unittests. Default to false.
-export SKIP_NUM=12 # skip the first 12 instances from the dataset
-```
-
-Following is the basic command to start the evaluation.
-
-You can update the arguments in the script
-`evaluation/benchmarks/aider_bench/scripts/run_infer.sh`, such as `--max-iterations`,
-`--eval-num-workers` and so on:
-
-- `--agent-cls`, the agent to use. For example, `CodeActAgent`.
-- `--llm-config`: the LLM configuration to use. For example, `eval_gpt4_1106_preview`.
-- `--max-iterations`: the max allowed number of iterations to run the evaluation. Default: `30`.
-- `--eval-num-workers`: the number of workers to use for evaluation. Default: `1`.
-- `--eval-n-limit`: the number of examples to evaluate. For example, `100`.
-- `--eval-ids`: the IDs of the examples to evaluate (comma separated). For example, `"1,3,10"`.
-
-```bash
-./evaluation/benchmarks/aider_bench/scripts/run_infer.sh eval_gpt35_turbo HEAD CodeActAgent 100 1 "1,3,10"
-```
-
-### Run Inference on `RemoteRuntime`
-
-This is in beta. Fill out [this form](https://docs.google.com/forms/d/e/1FAIpQLSckVz_JFwg2_mOxNZjCtr7aoBFI2Mwdan3f75J_TrdMS1JV2g/viewform) to apply if you want to try this out!
-
-
-```bash
-./evaluation/benchmarks/aider_bench/scripts/run_infer.sh [model_config] [git-version] [agent] [eval_limit] [eval-num-workers] [eval_ids]
-
-# Example - This runs evaluation on CodeActAgent for 133 instances on aider_bench test set, with 2 workers running in parallel
-export ALLHANDS_API_KEY="YOUR-API-KEY"
-export RUNTIME=remote
-export SANDBOX_REMOTE_RUNTIME_API_URL="https://runtime.eval.all-hands.dev"
-./evaluation/benchmarks/aider_bench/scripts/run_infer.sh llm.eval HEAD CodeActAgent 133 2
-```
-
-## Summarize Results
-
-```bash
-poetry run python ./evaluation/benchmarks/aider_bench/scripts/summarize_results.py [path_to_output_jsonl_file]
-```
-
-Full example:
-
-```bash
-poetry run python ./evaluation/benchmarks/aider_bench/scripts/summarize_results.py evaluation/evaluation_outputs/outputs/AiderBench/CodeActAgent/claude-3-5-sonnet@20240620_maxiter_30_N_v1.9/output.jsonl
-```
-
-This will list the instances that passed and the instances that failed. For each
-instance, the corresponding set of test cases (which can vary for each instance)
-are run on the file edited by the agent. We consider an instance to be passed
-only if ALL test cases are passed. Sometimes even a single failed test case will
-cause the entire instance to be marked as failed.
-
-You can inspect the `test_results` field in the `output.jsonl` file to find the exact
-outcome of the tests. If there are no syntax or indentation errors, you can
-expect to see something like "`..F...EF..`", where "`.`" means the test case
-passed, "`E`" means there was an error while executing the test case and "`F`"
-means some assertion failed and some returned output was not as expected.

evaluation/benchmarks/aider_bench/create_dataset.py

@@ -1,47 +0,0 @@
-# This file was used to create the hugging face dataset from the exercism/python
-# github repo.
-# Refer to: https://github.com/exercism/python/tree/main/exercises/practice
-
-import os
-from pathlib import Path
-
-from datasets import Dataset
-
-tests = sorted(os.listdir('practice/'))
-dataset = {
-    'instance_id': [],
-    'instance_name': [],
-    'instruction': [],
-    'signature': [],
-    'test': [],
-}
-
-for i, test in enumerate(tests):
-    testdir = Path(f'practice/{test}/')
-
-    dataset['instance_id'].append(i)
-    dataset['instance_name'].append(testdir.name.replace('-', '_'))
-
-    # if len(glob.glob(f'practice/{testdir.name}/*.py')) != 2:
-    #     print(testdir.name)
-
-    instructions = ''
-    introduction = testdir / '.docs/introduction.md'
-    if introduction.exists():
-        instructions += introduction.read_text()
-    instructions += (testdir / '.docs/instructions.md').read_text()
-    instructions_append = testdir / '.docs/instructions.append.md'
-    if instructions_append.exists():
-        instructions += instructions_append.read_text()
-
-    dataset['instruction'].append(instructions)
-
-    signature_file = testdir / (testdir.name + '.py').replace('-', '_')
-    dataset['signature'].append(signature_file.read_text())
-
-    test_file = testdir / (testdir.name + '_test.py').replace('-', '_')
-    dataset['test'].append(test_file.read_text())
-
-ds = Dataset.from_dict(dataset)
-
-ds.push_to_hub('RajMaheshwari/Exercism-Python')

evaluation/benchmarks/aider_bench/helper.py

@@ -1,22 +0,0 @@
-from evaluation.utils.shared import codeact_user_response
-
-INSTRUCTIONS_ADDENDUM = """
-####
-
-Use the above instructions to modify the supplied files: {signature_file}
-Don't change the names of existing functions or classes, as they may be referenced from other code like unit tests, etc.
-
-Only use standard python libraries, don't suggest installing any packages.
-"""
-
-
-FAKE_RESPONSES = {
-    'CodeActAgent': codeact_user_response,
-}
-
-INST_SUFFIXES: dict[str, str] = {
-    'CodeActAgent': (
-        'REMEMBER: All edits must be made directly in the files. Do NOT send'
-        ' the edited file as output to the user.\n'
-    )
-}

evaluation/benchmarks/aider_bench/run_infer.py

@@ -1,306 +0,0 @@
-import asyncio
-import copy
-import os
-import tempfile
-from typing import Any
-
-import pandas as pd
-from datasets import load_dataset
-
-from evaluation.benchmarks.aider_bench.helper import (
-    FAKE_RESPONSES,
-    INST_SUFFIXES,
-    INSTRUCTIONS_ADDENDUM,
-)
-from evaluation.utils.shared import (
-    EvalMetadata,
-    EvalOutput,
-    compatibility_for_eval_history_pairs,
-    get_default_sandbox_config_for_eval,
-    get_metrics,
-    get_openhands_config_for_eval,
-    make_metadata,
-    prepare_dataset,
-    reset_logger_for_multiprocessing,
-    run_evaluation,
-)
-from openhands.controller.state.state import State
-from openhands.core.config import (
-    OpenHandsConfig,
-    get_llm_config_arg,
-    load_from_toml,
-    parse_arguments,
-)
-from openhands.core.logger import openhands_logger as logger
-from openhands.core.main import create_runtime, run_controller
-from openhands.events.action import CmdRunAction, MessageAction
-from openhands.events.observation import CmdOutputObservation
-from openhands.runtime.base import Runtime
-from openhands.utils.async_utils import call_async_from_sync
-
-# Configure visibility of unit tests to the Agent.
-USE_UNIT_TESTS = os.environ.get('USE_UNIT_TESTS', 'false').lower() == 'true'
-SKIP_NUM = os.environ.get('SKIP_NUM')
-SKIP_NUM = (
-    int(SKIP_NUM) if SKIP_NUM and SKIP_NUM.isdigit() and int(SKIP_NUM) >= 0 else None
-)
-
-
-def get_config(
-    metadata: EvalMetadata,
-) -> OpenHandsConfig:
-    sandbox_config = get_default_sandbox_config_for_eval()
-    sandbox_config.base_container_image = 'python:3.11-bookworm'
-    config = get_openhands_config_for_eval(
-        metadata=metadata,
-        sandbox_config=sandbox_config,
-        runtime=os.environ.get('RUNTIME', 'docker'),
-    )
-    config.set_llm_config(metadata.llm_config)
-    agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.enable_prompt_extensions = False
-
-    # copy 'draft_editor' config if exists
-    config_copy = copy.deepcopy(config)
-    load_from_toml(config_copy)
-    if 'draft_editor' in config_copy.llms:
-        config.set_llm_config(config_copy.llms['draft_editor'], 'draft_editor')
-
-    return config
-
-
-def initialize_runtime(
-    runtime: Runtime,
-    instance: pd.Series,
-):
-    """Initialize the runtime for the agent.
-
-    This function is called before the runtime is used to run the agent.
-    """
-    logger.info(f'\n{"-" * 50} BEGIN Runtime Initialization Fn {"-" * 50}\n')
-    obs: CmdOutputObservation
-
-    # Set instance id
-    action = CmdRunAction(command='mkdir -p /workspace')
-    logger.info(action, extra={'msg_type': 'ACTION'})
-    obs = runtime.run_action(action)
-    assert obs.exit_code == 0
-
-    action = CmdRunAction(command='cd /workspace')
-    logger.info(action, extra={'msg_type': 'ACTION'})
-    obs = runtime.run_action(action)
-    assert obs.exit_code == 0
-
-    with tempfile.TemporaryDirectory() as tmpdir:
-        file_path = os.path.join(tmpdir, f'{instance.instance_name}.py')
-        with open(file_path, 'w') as f:
-            f.write(instance.signature)
-        runtime.copy_to(
-            file_path,
-            '/workspace',
-        )
-        if USE_UNIT_TESTS:
-            file_path = os.path.join(tmpdir, f'{instance.instance_name}_test.py')
-            with open(file_path, 'w') as f:
-                f.write(instance.test)
-            runtime.copy_to(
-                file_path,
-                '/workspace',
-            )
-    logger.info(f'\n{"-" * 50} END Runtime Initialization Fn {"-" * 50}\n')
-
-
-def complete_runtime(
-    runtime: Runtime,
-    instance: pd.Series,
-) -> dict[str, Any]:
-    """Complete the runtime for the agent.
-
-    This function is called before the runtime is used to run the agent.
-    If you need to do something in the sandbox to get the correctness metric after
-    the agent has run, modify this function.
-    """
-    logger.info(f'\n{"-" * 50} BEGIN Runtime Completion Fn {"-" * 50}\n')
-    obs: CmdOutputObservation
-
-    # Rewriting the test file to ignore any changes Agent may have made.
-    script_name = f'{instance.instance_name}_test.py'
-    with tempfile.TemporaryDirectory() as tmpdir:
-        file_path = os.path.join(tmpdir, script_name)
-        with open(file_path, 'w') as f:
-            f.write(instance.test)
-        runtime.copy_to(
-            file_path,
-            '/workspace',
-        )
-        logger.info(f'Running test file: {script_name}')
-
-    action = CmdRunAction(command=f'python3 -m unittest {script_name}')
-    logger.info(action, extra={'msg_type': 'ACTION'})
-    obs = runtime.run_action(action)
-    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
-
-    exit_code = 1
-    if isinstance(obs, CmdOutputObservation):
-        exit_code = obs.exit_code
-
-    logger.info(f'\n{"-" * 50} END Runtime Completion Fn {"-" * 50}\n')
-
-    runtime.close()
-
-    return {
-        'test_output': obs.content,
-        'exit_code': exit_code,
-    }
-
-
-def process_instance(
-    instance: pd.Series,
-    metadata: EvalMetadata,
-    reset_logger: bool = True,
-) -> EvalOutput:
-    config = get_config(metadata)
-
-    # Setup the logger properly, so you can run multi-processing to parallelize the evaluation
-    if reset_logger:
-        log_dir = os.path.join(metadata.eval_output_dir, 'infer_logs')
-        reset_logger_for_multiprocessing(logger, str(instance.instance_id), log_dir)
-    else:
-        logger.info(
-            f'\nStarting evaluation for instance {str(instance.instance_id)}.\n'
-        )
-
-    # =============================================
-    # build instruction
-    # =============================================
-
-    # Prepare instruction
-    logger.info(instance)
-    instruction = instance.instruction
-    instruction += INSTRUCTIONS_ADDENDUM.format(
-        signature_file=f'{instance.instance_name}.py',
-    )
-    if USE_UNIT_TESTS:
-        logger.info(
-            f'\nInstruction to run test_file: {instance.instance_name}_test.py\n'
-        )
-        instruction += (
-            f'Use `python -m unittest {instance.instance_name}_test.py` to run the test_file '
-            'and verify the correctness of your solution. DO NOT EDIT the test file.\n\n'
-        )
-
-    instruction += (
-        'IMPORTANT: You should ONLY interact with the environment provided '
-        'to you AND NEVER ASK FOR HUMAN HELP.\n'
-    )
-    # NOTE: You can actually set slightly different instruction for different agents
-    instruction += INST_SUFFIXES[metadata.agent_class]
-
-    # =============================================
-    # create sandbox and run the agent
-    # =============================================
-
-    runtime: Runtime = create_runtime(config)
-    call_async_from_sync(runtime.connect)
-    initialize_runtime(runtime, instance=instance)
-
-    # Here's how you can run the agent (similar to the `main` function) and get the final task state
-    state: State | None = asyncio.run(
-        run_controller(
-            config=config,
-            initial_user_action=MessageAction(content=instruction),
-            runtime=runtime,
-            fake_user_response_fn=FAKE_RESPONSES[metadata.agent_class],
-        )
-    )
-    if state is None:
-        raise ValueError('State should not be None.')
-
-    # # =============================================
-    # # result evaluation
-    # # =============================================
-
-    return_val = complete_runtime(runtime, instance)
-    exit_code = return_val['exit_code']
-    test_output = return_val['test_output']
-
-    errors = []
-    test_cases = None
-    if test_output.find('SyntaxError') != -1:
-        errors += 'SyntaxError'
-    elif test_output.find('IndentationError') != -1:
-        errors += 'IndentationError'
-    else:
-        test_cases = test_output[: test_output.find('\r')]
-
-    test_result = {
-        'exit_code': exit_code,
-        'test_cases': test_cases,
-        'errors': errors,
-    }
-
-    # history is now available as a stream of events, rather than list of pairs of (Action, Observation)
-    # for compatibility with the existing output format, we can remake the pairs here
-    # remove when it becomes unnecessary
-    histories = compatibility_for_eval_history_pairs(state.history)
-    metrics = get_metrics(state)
-
-    # Save the output
-    output = EvalOutput(
-        instance_id=str(instance.instance_id),
-        instance=instance.to_dict(),
-        instruction=instruction,
-        metadata=metadata,
-        history=histories,
-        metrics=metrics,
-        error=state.last_error if state and state.last_error else None,
-        test_result=test_result,
-    )
-    return output
-
-
-if __name__ == '__main__':
-    args = parse_arguments()
-    dataset = load_dataset('RajMaheshwari/Exercism-Python')
-    aider_bench_tests = dataset['train'].to_pandas()
-
-    llm_config = None
-    if args.llm_config:
-        llm_config = get_llm_config_arg(args.llm_config)
-        # modify_params must be False for evaluation purpose, for reproducibility and accurancy of results
-        llm_config.modify_params = False
-
-    if llm_config is None:
-        raise ValueError(f'Could not find LLM config: --llm_config {args.llm_config}')
-
-    metadata = make_metadata(
-        llm_config,
-        'AiderBench',
-        args.agent_cls,
-        args.max_iterations,
-        args.eval_note,
-        args.eval_output_dir,
-    )
-    output_file = os.path.join(metadata.eval_output_dir, 'output.jsonl')
-
-    # Parse dataset IDs if provided
-    eval_ids = None
-    if args.eval_ids:
-        eval_ids = str(args.eval_ids).split(',')
-        logger.info(f'\nUsing specific dataset IDs: {eval_ids}\n')
-
-    instances = prepare_dataset(
-        aider_bench_tests,
-        output_file,
-        args.eval_n_limit,
-        eval_ids=eval_ids,
-        skip_num=SKIP_NUM,
-    )
-
-    run_evaluation(
-        instances,
-        metadata,
-        output_file,
-        args.eval_num_workers,
-        process_instance,
-    )

evaluation/benchmarks/aider_bench/scripts/run_infer.sh

@@ -1,60 +0,0 @@
-#!/usr/bin/env bash
-set -eo pipefail
-
-source "evaluation/utils/version_control.sh"
-
-MODEL_CONFIG=$1
-COMMIT_HASH=$2
-AGENT=$3
-EVAL_LIMIT=$4
-NUM_WORKERS=$5
-EVAL_IDS=$6
-
-if [ -z "$NUM_WORKERS" ]; then
-  NUM_WORKERS=1
-  echo "Number of workers not specified, use default $NUM_WORKERS"
-fi
-checkout_eval_branch
-
-if [ -z "$AGENT" ]; then
-  echo "Agent not specified, use default CodeActAgent"
-  AGENT="CodeActAgent"
-fi
-
-get_openhands_version
-
-echo "AGENT: $AGENT"
-echo "OPENHANDS_VERSION: $OPENHANDS_VERSION"
-echo "MODEL_CONFIG: $MODEL_CONFIG"
-
-EVAL_NOTE=$OPENHANDS_VERSION
-
-# Default to NOT use unit tests.
-if [ -z "$USE_UNIT_TESTS" ]; then
-  export USE_UNIT_TESTS=false
-fi
-echo "USE_UNIT_TESTS: $USE_UNIT_TESTS"
-# If use unit tests, set EVAL_NOTE to the commit hash
-if [ "$USE_UNIT_TESTS" = true ]; then
-  EVAL_NOTE=$EVAL_NOTE-w-test
-fi
-
-COMMAND="export PYTHONPATH=evaluation/benchmarks/aider_bench:\$PYTHONPATH && poetry run python evaluation/benchmarks/aider_bench/run_infer.py \
-  --agent-cls $AGENT \
-  --llm-config $MODEL_CONFIG \
-  --max-iterations 30 \
-  --eval-num-workers $NUM_WORKERS \
-  --eval-note $EVAL_NOTE"
-
-if [ -n "$EVAL_LIMIT" ]; then
-  echo "EVAL_LIMIT: $EVAL_LIMIT"
-  COMMAND="$COMMAND --eval-n-limit $EVAL_LIMIT"
-fi
-
-if [ -n "$EVAL_IDS" ]; then
-  echo "EVAL_IDS: $EVAL_IDS"
-  COMMAND="$COMMAND --eval-ids $EVAL_IDS"
-fi
-
-# Run the command
-eval $COMMAND

evaluation/benchmarks/aider_bench/scripts/summarize_results.py

@@ -1,70 +0,0 @@
-import argparse
-
-import numpy as np
-import pandas as pd
-
-
-def extract_test_results(df: pd.DataFrame) -> tuple[list[str], list[str]]:
-    passed = []
-    failed = []
-    for _, row in df.iterrows():
-        instance_id = row['instance_id']
-        resolved = False
-        if 'test_result' in row and 'exit_code' in row['test_result']:
-            resolved = row['test_result']['exit_code'] == 0
-        if resolved:
-            passed.append(instance_id)
-        else:
-            failed.append(instance_id)
-    return passed, failed
-
-
-def visualize_results(df: pd.DataFrame):
-    df1 = pd.DataFrame()
-    df1['cost'] = df['metrics'].apply(pd.Series)['accumulated_cost']
-    df1['result'] = (
-        df['test_result'].apply(pd.Series)['exit_code'].map({0: 'Pass', 1: 'Fail'})
-    )
-    df1['actions'] = pd.Series([len(a) - 1 for a in df['history']])
-
-    passed = np.sum(df1['result'] == 'Pass')
-    total = df.shape[0]
-    resolve_rate = round((passed / total) * 100, 2)
-
-    print('Number of passed tests:', f'{passed}/{total} {resolve_rate:.2f}%')
-    print('\nDescriptive statistics for number of actions:')
-    print(df1['actions'].describe())
-    print('\nDescriptive statistics for costs:')
-    print(df1['cost'].describe())
-
-    # Bin counts for actions
-    action_bins = pd.cut(df1['actions'], bins=range(0, 32, 2))
-    print('\nAction bin counts:')
-    print(action_bins.value_counts().sort_index())
-
-    # Bin counts for costs
-    cost_bins = pd.cut(df1['cost'], bins=10)
-    print('\nCost bin counts:')
-    print(cost_bins.value_counts().sort_index())
-
-    return resolve_rate
-
-
-if __name__ == '__main__':
-    parser = argparse.ArgumentParser(description='Summarize AiderBench results')
-    parser.add_argument('input_filepath', type=str, help='Path to the JSONL file')
-    args = parser.parse_args()
-
-    # Create DataFrame from JSONL file
-    df = pd.read_json(args.input_filepath, lines=True)
-
-    passed_tests, failed_tests = extract_test_results(df)
-    resolve_rate = visualize_results(df)
-
-    print(
-        f'\nPassed {len(passed_tests)} tests, failed {len(failed_tests)} tests, resolve rate = {resolve_rate:.2f}%'
-    )
-    print('PASSED TESTS:')
-    print(passed_tests)
-    print('FAILED TESTS:')
-    print(failed_tests)