switch to use --rss to restrict memory use in bash session

Signed-off-by: José Luis Di Biase <josx@interorganic.com.ar>
Co-authored-by: José Luis Di Biase <josx@interorganic.com.ar>
Co-authored-by: Oriana <oriana@camba.coop>
Co-authored-by: Charlie <charlie@camba.coop>
Co-authored-by: Juan Manuel Daza <61162223+juanmanueldaza@users.noreply.github.com>
Co-authored-by: Juan Manuel Daza <juandaza@camba.coop>
Co-authored-by: Cody Kociemba <cody@symbaventures.com>
Co-authored-by: Rohit Malhotra <rohitvinodmalhotra@gmail.com>
Co-authored-by: Engel Nyst <enyst@users.noreply.github.com>

.github/ISSUE_TEMPLATE/bug_template.yml

@@ -30,6 +30,7 @@ body:
       description: How are you running OpenHands?
       options:
         - Docker command in README
+        - GitHub resolver
         - Development workflow
         - app.all-hands.dev
         - Other

.github/workflows/dummy-agent-test.yml

@@ -19,20 +19,6 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
-      - name: Free Disk Space (Ubuntu)
-        uses: jlumbroso/free-disk-space@main
-        with:
-          # this might remove tools that are actually needed,
-          # if set to "true" but frees about 6 GB
-          tool-cache: true
-          # all of these default to true, but feel free to set to
-          # "false" if necessary for your workflow
-          android: true
-          dotnet: true
-          haskell: true
-          large-packages: true
-          docker-images: false
-          swap-storage: true
       - name: Set up Docker Buildx
         id: buildx
         uses: docker/setup-buildx-action@v3

.github/workflows/ghcr-build.yml

@@ -41,22 +41,8 @@ jobs:
     steps:
       - name: Checkout
         uses: actions/checkout@v4
-      - name: Free Disk Space (Ubuntu)
-        uses: jlumbroso/free-disk-space@main
-        with:
-          # this might remove tools that are actually needed,
-          # if set to "true" but frees about 6 GB
-          tool-cache: true
-          # all of these default to true, but feel free to set to
-          # "false" if necessary for your workflow
-          android: true
-          dotnet: true
-          haskell: true
-          large-packages: true
-          docker-images: false
-          swap-storage: true
       - name: Set up QEMU
-        uses: docker/setup-qemu-action@v3.2.0
+        uses: docker/setup-qemu-action@v3.4.0
         with:
           image: tonistiigi/binfmt:latest
       - name: Login to GHCR
@@ -104,22 +90,8 @@ jobs:
     steps:
       - name: Checkout
         uses: actions/checkout@v4
-      - name: Free Disk Space (Ubuntu)
-        uses: jlumbroso/free-disk-space@main
-        with:
-          # this might remove tools that are actually needed,
-          # if set to "true" but frees about 6 GB
-          tool-cache: true
-          # all of these default to true, but feel free to set to
-          # "false" if necessary for your workflow
-          android: true
-          dotnet: true
-          haskell: true
-          large-packages: true
-          docker-images: false
-          swap-storage: true
       - name: Set up QEMU
-        uses: docker/setup-qemu-action@v3.2.0
+        uses: docker/setup-qemu-action@v3.4.0
         with:
           image: tonistiigi/binfmt:latest
       - name: Login to GHCR
@@ -219,7 +191,7 @@ jobs:
             exit 1
           fi
 
-  # Run unit tests with the EventStream runtime Docker images as root
+  # Run unit tests with the Docker runtime Docker images as root
   test_runtime_root:
     name: RT Unit Tests (Root)
     needs: [ghcr_build_runtime]
@@ -230,20 +202,6 @@ jobs:
         base_image: ['nikolaik']
     steps:
       - uses: actions/checkout@v4
-      - name: Free Disk Space (Ubuntu)
-        uses: jlumbroso/free-disk-space@main
-        with:
-          # this might remove tools that are actually needed,
-          # if set to "true" but frees about 6 GB
-          tool-cache: true
-          # all of these default to true, but feel free to set to
-          # "false" if necessary for your workflow
-          android: true
-          dotnet: true
-          haskell: true
-          large-packages: true
-          docker-images: false
-          swap-storage: true
       - name: Set up Docker Buildx
         id: buildx
         uses: docker/setup-buildx-action@v3
@@ -275,7 +233,7 @@ jobs:
         run: pipx install poetry
       - name: Install Python dependencies using Poetry
         run: make install-python-dependencies
-      - name: Run runtime tests
+      - name: Run docker runtime tests
         run: |
           # We install pytest-xdist in order to run tests across CPUs
           poetry run pip install pytest-xdist
@@ -286,7 +244,7 @@ jobs:
           image_name=ghcr.io/${{ github.repository_owner }}/runtime:${{ env.RELEVANT_SHA }}-${{ matrix.base_image }}
           image_name=$(echo $image_name | tr '[:upper:]' '[:lower:]')
 
-          TEST_RUNTIME=eventstream \
+          TEST_RUNTIME=docker \
           SANDBOX_USER_ID=$(id -u) \
           SANDBOX_RUNTIME_CONTAINER_IMAGE=$image_name \
           TEST_IN_CI=true \
@@ -297,7 +255,7 @@ jobs:
         env:
           CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}
 
-  # Run unit tests with the EventStream runtime Docker images as openhands user
+  # Run unit tests with the Docker runtime Docker images as openhands user
   test_runtime_oh:
     name: RT Unit Tests (openhands)
     runs-on: ubuntu-latest
@@ -307,20 +265,6 @@ jobs:
         base_image: ['nikolaik']
     steps:
       - uses: actions/checkout@v4
-      - name: Free Disk Space (Ubuntu)
-        uses: jlumbroso/free-disk-space@main
-        with:
-          # this might remove tools that are actually needed,
-          # if set to "true" but frees about 6 GB
-          tool-cache: true
-          # all of these default to true, but feel free to set to
-          # "false" if necessary for your workflow
-          android: true
-          dotnet: true
-          haskell: true
-          large-packages: true
-          docker-images: false
-          swap-storage: true
       - name: Set up Docker Buildx
         id: buildx
         uses: docker/setup-buildx-action@v3
@@ -363,7 +307,7 @@ jobs:
           image_name=ghcr.io/${{ github.repository_owner }}/runtime:${{ env.RELEVANT_SHA }}-${{ matrix.base_image }}
           image_name=$(echo $image_name | tr '[:upper:]' '[:lower:]')
 
-          TEST_RUNTIME=eventstream \
+          TEST_RUNTIME=docker \
           SANDBOX_USER_ID=$(id -u) \
           SANDBOX_RUNTIME_CONTAINER_IMAGE=$image_name \
           TEST_IN_CI=true \

.github/workflows/integration-runner.yml

@@ -56,6 +56,7 @@ jobs:
           LLM_MODEL: "litellm_proxy/claude-3-5-haiku-20241022"
           LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
           LLM_BASE_URL: ${{ secrets.LLM_BASE_URL }}
+          MAX_ITERATIONS: 10
         run: |
           echo "[llm.eval]" > config.toml
           echo "model = \"$LLM_MODEL\"" >> config.toml
@@ -70,7 +71,7 @@ jobs:
         env:
           SANDBOX_FORCE_REBUILD_RUNTIME: True
         run: |
-          poetry run ./evaluation/integration_tests/scripts/run_infer.sh llm.eval HEAD CodeActAgent '' $N_PROCESSES '' 'haiku_run'
+          poetry run ./evaluation/integration_tests/scripts/run_infer.sh llm.eval HEAD CodeActAgent '' 10 $N_PROCESSES '' 'haiku_run'
 
           # get integration tests report
           REPORT_FILE_HAIKU=$(find evaluation/evaluation_outputs/outputs/integration_tests/CodeActAgent/*haiku*_maxiter_10_N* -name "report.md" -type f | head -n 1)
@@ -88,6 +89,7 @@ jobs:
           LLM_MODEL: "litellm_proxy/deepseek-chat"
           LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
           LLM_BASE_URL: ${{ secrets.LLM_BASE_URL }}
+          MAX_ITERATIONS: 10
         run: |
           echo "[llm.eval]" > config.toml
           echo "model = \"$LLM_MODEL\"" >> config.toml
@@ -99,7 +101,7 @@ jobs:
         env:
           SANDBOX_FORCE_REBUILD_RUNTIME: True
         run: |
-          poetry run ./evaluation/integration_tests/scripts/run_infer.sh llm.eval HEAD CodeActAgent '' $N_PROCESSES '' 'deepseek_run'
+          poetry run ./evaluation/integration_tests/scripts/run_infer.sh llm.eval HEAD CodeActAgent '' 10 $N_PROCESSES '' 'deepseek_run'
 
           # get integration tests report
           REPORT_FILE_DEEPSEEK=$(find evaluation/evaluation_outputs/outputs/integration_tests/CodeActAgent/deepseek*_maxiter_10_N* -name "report.md" -type f | head -n 1)
@@ -109,11 +111,104 @@ jobs:
           echo >> $GITHUB_ENV
           echo "EOF" >> $GITHUB_ENV
 
+      # -------------------------------------------------------------
+      # Run DelegatorAgent tests for Haiku, limited to t01 and t02
+      - name: Wait a little bit (again)
+        run: sleep 5
+
+      - name: Configure config.toml for testing DelegatorAgent (Haiku)
+        env:
+          LLM_MODEL: "litellm_proxy/claude-3-5-haiku-20241022"
+          LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
+          LLM_BASE_URL: ${{ secrets.LLM_BASE_URL }}
+          MAX_ITERATIONS: 30
+        run: |
+          echo "[llm.eval]" > config.toml
+          echo "model = \"$LLM_MODEL\"" >> config.toml
+          echo "api_key = \"$LLM_API_KEY\"" >> config.toml
+          echo "base_url = \"$LLM_BASE_URL\"" >> config.toml
+          echo "temperature = 0.0" >> config.toml
+
+      - name: Run integration test evaluation for DelegatorAgent (Haiku)
+        env:
+          SANDBOX_FORCE_REBUILD_RUNTIME: True
+        run: |
+          poetry run ./evaluation/integration_tests/scripts/run_infer.sh llm.eval HEAD DelegatorAgent '' 30 $N_PROCESSES "t01_fix_simple_typo,t02_add_bash_hello" 'delegator_haiku_run'
+
+          # Find and export the delegator test results
+          REPORT_FILE_DELEGATOR_HAIKU=$(find evaluation/evaluation_outputs/outputs/integration_tests/DelegatorAgent/*haiku*_maxiter_30_N* -name "report.md" -type f | head -n 1)
+          echo "REPORT_FILE_DELEGATOR_HAIKU: $REPORT_FILE_DELEGATOR_HAIKU"
+          echo "INTEGRATION_TEST_REPORT_DELEGATOR_HAIKU<<EOF" >> $GITHUB_ENV
+          cat $REPORT_FILE_DELEGATOR_HAIKU >> $GITHUB_ENV
+          echo >> $GITHUB_ENV
+          echo "EOF" >> $GITHUB_ENV
+
+      # -------------------------------------------------------------
+      # Run DelegatorAgent tests for DeepSeek, limited to t01 and t02
+      - name: Wait a little bit (again)
+        run: sleep 5
+
+      - name: Configure config.toml for testing DelegatorAgent (DeepSeek)
+        env:
+          LLM_MODEL: "litellm_proxy/deepseek-chat"
+          LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
+          LLM_BASE_URL: ${{ secrets.LLM_BASE_URL }}
+          MAX_ITERATIONS: 30
+        run: |
+          echo "[llm.eval]" > config.toml
+          echo "model = \"$LLM_MODEL\"" >> config.toml
+          echo "api_key = \"$LLM_API_KEY\"" >> config.toml
+          echo "base_url = \"$LLM_BASE_URL\"" >> config.toml
+          echo "temperature = 0.0" >> config.toml
+      - name: Run integration test evaluation for DelegatorAgent (DeepSeek)
+        env:
+          SANDBOX_FORCE_REBUILD_RUNTIME: True
+        run: |
+          poetry run ./evaluation/integration_tests/scripts/run_infer.sh llm.eval HEAD DelegatorAgent '' 30 $N_PROCESSES "t01_fix_simple_typo,t02_add_bash_hello" 'delegator_deepseek_run'
+
+          # Find and export the delegator test results
+          REPORT_FILE_DELEGATOR_DEEPSEEK=$(find evaluation/evaluation_outputs/outputs/integration_tests/DelegatorAgent/deepseek*_maxiter_30_N* -name "report.md" -type f | head -n 1)
+          echo "REPORT_FILE_DELEGATOR_DEEPSEEK: $REPORT_FILE_DELEGATOR_DEEPSEEK"
+          echo "INTEGRATION_TEST_REPORT_DELEGATOR_DEEPSEEK<<EOF" >> $GITHUB_ENV
+          cat $REPORT_FILE_DELEGATOR_DEEPSEEK >> $GITHUB_ENV
+          echo >> $GITHUB_ENV
+          echo "EOF" >> $GITHUB_ENV
+      # -------------------------------------------------------------
+      # Run VisualBrowsingAgent tests for DeepSeek, limited to t05 and t06
+      - name: Wait a little bit (again)
+        run: sleep 5
+
+      - name: Configure config.toml for testing VisualBrowsingAgent (DeepSeek)
+        env:
+          LLM_MODEL: "litellm_proxy/deepseek-chat"
+          LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
+          LLM_BASE_URL: ${{ secrets.LLM_BASE_URL }}
+          MAX_ITERATIONS: 15
+        run: |
+          echo "[llm.eval]" > config.toml
+          echo "model = \"$LLM_MODEL\"" >> config.toml
+          echo "api_key = \"$LLM_API_KEY\"" >> config.toml
+          echo "base_url = \"$LLM_BASE_URL\"" >> config.toml
+          echo "temperature = 0.0" >> config.toml
+      - name: Run integration test evaluation for VisualBrowsingAgent (DeepSeek)
+        env:
+          SANDBOX_FORCE_REBUILD_RUNTIME: True
+        run: |
+          poetry run ./evaluation/integration_tests/scripts/run_infer.sh llm.eval HEAD VisualBrowsingAgent '' 15 $N_PROCESSES "t05_simple_browsing,t06_github_pr_browsing.py" 'visualbrowsing_deepseek_run'
+
+          # Find and export the visual browsing agent test results
+          REPORT_FILE_VISUALBROWSING_DEEPSEEK=$(find evaluation/evaluation_outputs/outputs/integration_tests/VisualBrowsingAgent/deepseek*_maxiter_15_N* -name "report.md" -type f | head -n 1)
+          echo "REPORT_FILE_VISUALBROWSING_DEEPSEEK: $REPORT_FILE_VISUALBROWSING_DEEPSEEK"
+          echo "INTEGRATION_TEST_REPORT_VISUALBROWSING_DEEPSEEK<<EOF" >> $GITHUB_ENV
+          cat $REPORT_FILE_VISUALBROWSING_DEEPSEEK >> $GITHUB_ENV
+          echo >> $GITHUB_ENV
+          echo "EOF" >> $GITHUB_ENV
+
       - name: Create archive of evaluation outputs
         run: |
           TIMESTAMP=$(date +'%y-%m-%d-%H-%M')
           cd evaluation/evaluation_outputs/outputs  # Change to the outputs directory
-          tar -czvf ../../../integration_tests_${TIMESTAMP}.tar.gz integration_tests/CodeActAgent/*  # Only include the actual result directories
+          tar -czvf ../../../integration_tests_${TIMESTAMP}.tar.gz integration_tests/CodeActAgent/* integration_tests/DelegatorAgent/* integration_tests/VisualBrowsingAgent/* # Only include the actual result directories
 
       - name: Upload evaluation results as artifact
         uses: actions/upload-artifact@v4
@@ -154,5 +249,14 @@ jobs:
               **Integration Tests Report (DeepSeek)**
               DeepSeek LLM Test Results:
               ${{ env.INTEGRATION_TEST_REPORT_DEEPSEEK }}
+              ---
+                **Integration Tests Report Delegator (Haiku)**
+              ${{ env.INTEGRATION_TEST_REPORT_DELEGATOR_HAIKU }}
+              ---
+              **Integration Tests Report Delegator (DeepSeek)**
+              ${{ env.INTEGRATION_TEST_REPORT_DELEGATOR_DEEPSEEK }}
+              ---
+              **Integration Tests Report VisualBrowsing (DeepSeek)**
+              ${{ env.INTEGRATION_TEST_REPORT_VISUALBROWSING_DEEPSEEK }}
               ---
               Download testing outputs (includes both Haiku and DeepSeek results): [Download](${{ steps.upload_results_artifact.outputs.artifact-url }})

.github/workflows/openhands-resolver.yml

@@ -20,6 +20,10 @@ on:
         required: false
         type: string
         default: "anthropic/claude-3-5-sonnet-20241022"
+      LLM_API_VERSION:
+        required: false
+        type: string
+        default: ""
       base_container_image:
         required: false
         type: string
@@ -84,6 +88,10 @@ jobs:
         run: |
           python -m pip index versions openhands-ai > openhands_versions.txt
           OPENHANDS_VERSION=$(head -n 1 openhands_versions.txt | awk '{print $2}' | tr -d '()')
+          # Ensure requirements.txt ends with newline before appending
+          if [ -f requirements.txt ] && [ -s requirements.txt ]; then
+            sed -i -e '$a\' requirements.txt
+          fi
           echo "openhands-ai==${OPENHANDS_VERSION}" >> requirements.txt
           cat requirements.txt
 
@@ -112,6 +120,7 @@ jobs:
           LLM_MODEL: ${{ secrets.LLM_MODEL || inputs.LLM_MODEL }}
           LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
           LLM_BASE_URL: ${{ secrets.LLM_BASE_URL }}
+          LLM_API_VERSION: ${{ inputs.LLM_API_VERSION }}
           PAT_TOKEN: ${{ secrets.PAT_TOKEN }}
           PAT_USERNAME: ${{ secrets.PAT_USERNAME }}
           GITHUB_TOKEN: ${{ github.token }}
@@ -168,7 +177,7 @@ jobs:
           echo "SANDBOX_ENV_BASE_CONTAINER_IMAGE=${{ inputs.base_container_image }}" >> $GITHUB_ENV
 
           # Set branch variables
-          echo "TARGET_BRANCH=${{ inputs.target_branch }}" >> $GITHUB_ENV
+          echo "TARGET_BRANCH=${{ inputs.target_branch || 'main' }}" >> $GITHUB_ENV
 
       - name: Comment on issue with start message
         uses: actions/github-script@v7
@@ -184,6 +193,7 @@ jobs:
             });
 
       - name: Install OpenHands
+        id: install_openhands
         uses: actions/github-script@v7
         env:
           COMMENT_BODY: ${{ github.event.comment.body || '' }}
@@ -196,7 +206,6 @@ jobs:
             const reviewBody = process.env.REVIEW_BODY.trim();
             const labelName = process.env.LABEL_NAME.trim();
             const eventName = process.env.EVENT_NAME.trim();
-
             // Check conditions
             const isExperimentalLabel = labelName === "fix-me-experimental";
             const isIssueCommentExperimental =
@@ -205,6 +214,9 @@ jobs:
             const isReviewCommentExperimental =
               eventName === "pull_request_review" && reviewBody.includes("@openhands-agent-exp");
 
+            // Set output variable
+            core.setOutput('isExperimental', isExperimentalLabel || isIssueCommentExperimental || isReviewCommentExperimental);
+
             // Perform package installation
             if (isExperimentalLabel || isIssueCommentExperimental || isReviewCommentExperimental) {
               console.log("Installing experimental OpenHands...");
@@ -223,6 +235,7 @@ jobs:
           LLM_MODEL: ${{ secrets.LLM_MODEL || inputs.LLM_MODEL }}
           LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
           LLM_BASE_URL: ${{ secrets.LLM_BASE_URL }}
+          LLM_API_VERSION: ${{ inputs.LLM_API_VERSION }}
           PYTHONPATH: ""
         run: |
           cd /tmp && python -m openhands.resolver.resolve_issue \
@@ -230,7 +243,8 @@ jobs:
             --issue-number ${{ env.ISSUE_NUMBER }} \
             --issue-type ${{ env.ISSUE_TYPE }} \
             --max-iterations ${{ env.MAX_ITERATIONS }} \
-            --comment-id ${{ env.COMMENT_ID }}
+            --comment-id ${{ env.COMMENT_ID }} \
+            --is-experimental ${{ steps.install_openhands.outputs.isExperimental }}
 
       - name: Check resolution result
         id: check_result
@@ -257,11 +271,13 @@ jobs:
           LLM_MODEL: ${{ secrets.LLM_MODEL || inputs.LLM_MODEL }}
           LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
           LLM_BASE_URL: ${{ secrets.LLM_BASE_URL }}
+          LLM_API_VERSION: ${{ inputs.LLM_API_VERSION }}
           PYTHONPATH: ""
         run: |
           if [ "${{ steps.check_result.outputs.RESOLUTION_SUCCESS }}" == "true" ]; then
             cd /tmp && python -m openhands.resolver.send_pull_request \
               --issue-number ${{ env.ISSUE_NUMBER }} \
+              --target-branch ${{ env.TARGET_BRANCH }} \
               --pr-type draft \
               --reviewer ${{ github.actor }} | tee pr_result.txt && \
               grep "draft created" pr_result.txt | sed 's/.*\///g' > pr_number.txt

.github/workflows/py-unit-tests-mac.yml

@@ -1,98 +0,0 @@
-# Workflow that runs python unit tests on mac
-name: Run Python Unit Tests Mac
-
-# This job is flaky so only run it nightly
-on:
-  schedule:
-    - cron: '0 0 * * *'
-
-jobs:
-  # Run python unit tests on macOS
-  test-on-macos:
-    name: Python Unit Tests on macOS
-    runs-on: macos-14
-    env:
-      INSTALL_DOCKER: '1' # Set to '0' to skip Docker installation
-    strategy:
-      matrix:
-        python-version: ['3.12']
-    steps:
-      - uses: actions/checkout@v4
-      - name: Set up Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v5
-        with:
-          python-version: ${{ matrix.python-version }}
-      - name: Cache Poetry dependencies
-        uses: actions/cache@v4
-        with:
-          path: |
-            ~/.cache/pypoetry
-            ~/.virtualenvs
-          key: ${{ runner.os }}-poetry-${{ hashFiles('**/poetry.lock') }}
-          restore-keys: |
-            ${{ runner.os }}-poetry-
-      - name: Install tmux
-        run: brew install tmux
-      - name: Install poetry via pipx
-        run: pipx install poetry
-      - name: Install Python dependencies using Poetry
-        run: poetry install --without evaluation,llama-index
-      - name: Install & Start Docker
-        if: env.INSTALL_DOCKER == '1'
-        run: |
-          INSTANCE_NAME="colima-${GITHUB_RUN_ID}"
-
-          # Uninstall colima to upgrade to the latest version
-          if brew list colima &>/dev/null; then
-            brew uninstall colima
-            # unlinking colima dependency: go
-            brew uninstall go@1.21
-          fi
-          rm -rf ~/.colima ~/.lima
-          brew install --HEAD colima
-          brew install docker
-
-          start_colima() {
-            # Find a free port in the range 10000-20000
-            RANDOM_PORT=$((RANDOM % 10001 + 10000))
-
-            # Original line:
-            if ! colima start --network-address --arch x86_64 --cpu=1 --memory=1 --verbose --ssh-port $RANDOM_PORT; then
-              echo "Failed to start Colima."
-              return 1
-            fi
-            return 0
-          }
-
-          # Attempt to start Colima for 5 total attempts:
-          ATTEMPT_LIMIT=5
-          for ((i=1; i<=ATTEMPT_LIMIT; i++)); do
-
-            if start_colima; then
-              echo "Colima started successfully."
-              break
-            else
-              colima stop -f
-              sleep 10
-              colima delete -f
-              if [ $i -eq $ATTEMPT_LIMIT ]; then
-                exit 1
-              fi
-              sleep 10
-            fi
-          done
-
-          # For testcontainers to find the Colima socket
-          # https://github.com/abiosoft/colima/blob/main/docs/FAQ.md#cannot-connect-to-the-docker-daemon-at-unixvarrundockersock-is-the-docker-daemon-running
-          sudo ln -sf $HOME/.colima/default/docker.sock /var/run/docker.sock
-      - name: Build Environment
-        run: make build
-      - name: Set up Docker Buildx
-        id: buildx
-        uses: docker/setup-buildx-action@v3
-      - name: Run Tests
-        run: poetry run pytest --forked --cov=openhands --cov-report=xml ./tests/unit --ignore=tests/unit/test_memory.py
-      - name: Upload coverage to Codecov
-        uses: codecov/codecov-action@v5
-        env:
-          CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}

.github/workflows/stale.yml

@@ -19,3 +19,4 @@ jobs:
           close-issue-message: 'This issue was closed because it has been stalled for over 30 days with no activity.'
           close-pr-message: 'This PR was closed because it has been stalled for over 30 days with no activity.'
           days-before-close: 7
+          operations-per-run: 150

.openhands/microagents/glossary.md

@@ -0,0 +1,172 @@
+# OpenHands Glossary
+
+### Agent
+The core AI entity in OpenHands that can perform software development tasks by interacting with tools, browsing the web, and modifying code.
+
+#### Agent Controller
+A component that manages the agent's lifecycle, handles its state, and coordinates interactions between the agent and various tools.
+
+#### Agent Delegation
+The ability of an agent to hand off specific tasks to other specialized agents for better task completion.
+
+#### Agent Hub
+A central registry of different agent types and their capabilities, allowing for easy agent selection and instantiation.
+
+#### Agent Skill
+A specific capability or function that an agent can perform, such as file manipulation, web browsing, or code editing.
+
+#### Agent State
+The current context and status of an agent, including its memory, active tools, and ongoing tasks.
+
+#### CodeAct Agent
+[A generalist agent in OpenHands](https://arxiv.org/abs/2407.16741) designed to perform tasks by editing and executing code.
+
+### Browser
+A system for web-based interactions and tasks.
+
+#### Browser Gym
+A testing and evaluation environment for browser-based agent interactions and tasks.
+
+#### Web Browser Tool
+A tool that enables agents to interact with web pages and perform web-based tasks.
+
+### Commands
+Terminal and execution related functionality.
+
+#### Bash Session
+A persistent terminal session that maintains state and history for bash command execution.
+This uses tmux under the hood.
+
+### Configuration
+System-wide settings and options.
+
+#### Agent Configuration
+Settings that define an agent's behavior, capabilities, and limitations, including available tools and runtime settings.
+
+#### Configuration Options
+Settings that control various aspects of OpenHands behavior, including runtime, security, and agent settings.
+
+#### LLM Config
+Configuration settings for language models used by agents, including model selection and parameters.
+
+#### LLM Draft Config
+Settings for draft mode operations with language models, typically used for faster, lower-quality responses.
+
+#### Runtime Configuration
+Settings that define how the runtime environment should be set up and operated.
+
+#### Security Options
+Configuration settings that control security features and restrictions.
+
+### Conversation
+A sequence of interactions between a user and an agent, including messages, actions, and their results.
+
+#### Conversation Info
+Metadata about a conversation, including its status, participants, and timeline.
+
+#### Conversation Manager
+A component that handles the creation, storage, and retrieval of conversations.
+
+#### Conversation Metadata
+Additional information about conversations, such as tags, timestamps, and related resources.
+
+#### Conversation Status
+The current state of a conversation, including whether it's active, completed, or failed.
+
+#### Conversation Store
+A storage system for maintaining conversation history and related data.
+
+### Events
+
+#### Event
+Every Conversation comprises a series of Events. Each Event is either an Action or an Observation.
+
+#### Event Stream
+A continuous flow of events that represents the ongoing activities and interactions in the system.
+
+#### Action
+A specific operation or command that an agent executes through available tools, such as running a command or editing a file.
+
+#### Observation
+The response or result returned by a tool after an agent's action, providing feedback about the action's outcome.
+
+### Interface
+Different ways to interact with OpenHands.
+
+#### CLI Mode
+A command-line interface mode for interacting with OpenHands agents without a graphical interface.
+
+#### GUI Mode
+A graphical user interface mode for interacting with OpenHands agents through a web interface.
+
+#### Headless Mode
+A mode of operation where OpenHands runs without a user interface, suitable for automation and scripting.
+
+### Agent Memory
+The system that decides which parts of the Event Stream (i.e. the conversation history) should be passed into each LLM prompt.
+
+#### Memory Store
+A storage system for maintaining agent memory and context across sessions.
+
+#### Condenser
+A component that processes and summarizes conversation history to maintain context while staying within token limits.
+
+#### Truncation
+A very simple Condenser strategy. Reduces conversation history or content to stay within token limits.
+
+### Microagent
+A specialized prompt that enhances OpenHands with domain-specific knowledge, repository-specific context, and task-specific workflows.
+
+#### Microagent Registry
+A central repository of available microagents and their configurations.
+
+#### Public Microagent
+A general-purpose microagent available to all OpenHands users, triggered by specific keywords.
+
+#### Repository Microagent
+A type of microagent that provides repository-specific context and guidelines, stored in the `.openhands/microagents/` directory.
+
+### Prompt
+Components for managing and processing prompts.
+
+#### Prompt Caching
+A system for caching and reusing common prompts to improve performance.
+
+#### Prompt Manager
+A component that handles the loading, processing, and management of prompts used by agents, including microagents.
+
+#### Response Parsing
+The process of interpreting and structuring responses from language models and tools.
+
+### Runtime
+The execution environment where agents perform their tasks, which can be local, remote, or containerized.
+
+#### Action Execution Server
+A REST API that receives agent actions (e.g. bash commands, python code, browsing actions), executes them in the runtime environment, and returns the results.
+
+#### Action Execution Client
+A component that handles the execution of actions in the runtime environment, managing the communication between the agent and the runtime.
+
+#### Docker Runtime
+A containerized runtime environment that provides isolation and reproducibility for agent operations.
+
+#### E2B Runtime
+A specialized runtime environment built on E2B for secure and isolated code execution.
+
+#### Local Runtime
+A runtime environment that executes on the local machine, suitable for development and testing.
+
+#### Modal Runtime
+A runtime environment built on Modal for scalable and distributed agent operations.
+
+#### Remote Runtime
+A sandboxed environment that executes code and commands remotely, providing isolation and security for agent operations.
+
+#### Runtime Builder
+A component that builds a Docker image for the Action Execution Server based on a user-specified base image.
+
+### Security
+Security-related components and features.
+
+#### Security Analyzer
+A component that checks agent actions for potential security risks.

CODE_OF_CONDUCT.md

@@ -18,24 +18,24 @@ diverse, inclusive, and healthy community.
 Examples of behavior that contributes to a positive environment for our
 community include:
 
-* Demonstrating empathy and kindness toward other people
-* Being respectful of differing opinions, viewpoints, and experiences
-* Giving and gracefully accepting constructive feedback
+* Demonstrating empathy and kindness toward other people.
+* Being respectful of differing opinions, viewpoints, and experiences.
+* Giving and gracefully accepting constructive feedback.
 * Accepting responsibility and apologizing to those affected by our mistakes,
-  and learning from the experience
+  and learning from the experience.
 * Focusing on what is best not just for us as individuals, but for the overall
-  community
+  community.
 
 Examples of unacceptable behavior include:
 
 * The use of sexualized language or imagery, and sexual attention or advances of
-  any kind
-* Trolling, insulting or derogatory comments, and personal or political attacks
-* Public or private harassment
+  any kind.
+* Trolling, insulting or derogatory comments, and personal or political attacks.
+* Public or private harassment.
 * Publishing others' private information, such as a physical or email address,
-  without their explicit permission
+  without their explicit permission.
 * Other conduct which could reasonably be considered inappropriate in a
-  professional setting
+  professional setting.
 
 ## Enforcement Responsibilities
 
@@ -61,7 +61,7 @@ representative at an online or offline event.
 
 Instances of abusive, harassing, or otherwise unacceptable behavior may be
 reported to the community leaders responsible for enforcement at
-contact@all-hands.dev
+contact@all-hands.dev.
 All complaints will be reviewed and investigated promptly and fairly.
 
 All community leaders are obligated to respect the privacy and security of the
@@ -113,6 +113,20 @@ individual, or aggression toward or disparagement of classes of individuals.
 **Consequence**: A permanent ban from any sort of public interaction within the
 community.
 
+### Slack and Discord Etiquettes
+
+These Slack and Discord etiquette guidelines are designed to foster an inclusive, respectful, and productive environment for all community members. By following these best practices, we ensure effective communication and collaboration while minimizing disruptions. Let’s work together to build a supportive and welcoming community!
+
+- Communicate respectfully and professionally, avoiding sarcasm or harsh language, and remember that tone can be difficult to interpret in text.
+- Use threads for specific discussions to keep channels organized and easier to follow.
+- Tag others only when their input is critical or urgent, and use @here, @channel or @everyone sparingly to minimize disruptions.
+- Be patient, as open-source contributors and maintainers often have other commitments and may need time to respond.
+- Post questions or discussions in the most relevant channel (e.g., for [slack - #general](https://app.slack.com/client/T06P212QSEA/C06P5NCGSFP) for general topics, [slack - #questions](https://openhands-ai.slack.com/archives/C06U8UTKSAD) for queries/questions, [discord - #general](https://discord.com/channels/1222935860639563850/1222935861386018885)).
+- When asking for help or raising issues, include necessary details like links, screenshots, or clear explanations to provide context.
+- Keep discussions in public channels whenever possible to allow others to benefit from the conversation, unless the matter is sensitive or private.
+- Always adhere to [our standards](https://github.com/All-Hands-AI/OpenHands/blob/main/CODE_OF_CONDUCT.md#our-standards) to ensure a welcoming and collaborative environment.
+- If you choose to mute a channel, consider setting up alerts for topics that still interest you to stay engaged. For Slack, Go to Settings → Notifications → My Keywords to add specific keywords that will notify you when mentioned. For example, if you're here for discussions about LLMs, mute the channel if it’s too busy, but set notifications to alert you only when “LLMs” appears in messages. Also for Discord, go to the channel notifications and choose the option that best describes your need.
+
 ## Attribution
 
 This Code of Conduct is adapted from the [Contributor Covenant][homepage],

CONTRIBUTING.md

@@ -11,11 +11,11 @@ To understand the codebase, please refer to the README in each module:
    - [agenthub](./openhands/agenthub/README.md)
    - [server](./openhands/server/README.md)
 
-## Setting up your development environment
+## Setting up Your Development Environment
 
 We have a separate doc [Development.md](https://github.com/All-Hands-AI/OpenHands/blob/main/Development.md) that tells you how to set up a development workflow.
 
-## How can I contribute?
+## How Can I Contribute?
 
 There are many ways that you can contribute:
 
@@ -23,7 +23,7 @@ There are many ways that you can contribute:
 2. **Send feedback** after each session by [clicking the thumbs-up thumbs-down buttons](https://docs.all-hands.dev/modules/usage/feedback), so we can see where things are working and failing, and also build an open dataset for training code agents.
 3. **Improve the Codebase** by sending [PRs](#sending-pull-requests-to-openhands) (see details below). In particular, we have some [good first issues](https://github.com/All-Hands-AI/OpenHands/labels/good%20first%20issue) that may be ones to start on.
 
-## What can I build?
+## What Can I Build?
 Here are a few ways you can help improve the codebase.
 
 #### UI/UX
@@ -35,7 +35,7 @@ of the application, please open an issue first, or better, join the #frontend ch
 to gather consensus from our design team first.
 
 #### Improving the agent
-Our main agent is the CodeAct agent. You can [see its prompts here](https://github.com/All-Hands-AI/OpenHands/tree/main/openhands/agenthub/codeact_agent)
+Our main agent is the CodeAct agent. You can [see its prompts here](https://github.com/All-Hands-AI/OpenHands/tree/main/openhands/agenthub/codeact_agent).
 
 Changes to these prompts, and to the underlying behavior in Python, can have a huge impact on user experience.
 You can try modifying the prompts to see how they change the behavior of the agent as you use the app
@@ -63,7 +63,7 @@ At the moment, we have two kinds of tests: [`unit`](./tests/unit) and [`integrat
 ## Sending Pull Requests to OpenHands
 
 You'll need to fork our repository to send us a Pull Request. You can learn more
-about how to fork a GitHub repo and open a PR with your changes in [this article](https://medium.com/swlh/forks-and-pull-requests-how-to-contribute-to-github-repos-8843fac34ce8)
+about how to fork a GitHub repo and open a PR with your changes in [this article](https://medium.com/swlh/forks-and-pull-requests-how-to-contribute-to-github-repos-8843fac34ce8).
 
 ### Pull Request title
 As described [here](https://github.com/commitizen/conventional-commit-types/blob/master/index.json), a valid PR title should begin with one of the following prefixes:
@@ -103,7 +103,7 @@ Further, if you see an issue you like, please leave a "thumbs-up" or a comment,
 
 ### Making Pull Requests
 
-We're generally happy to consider all [PRs](https://github.com/All-Hands-AI/OpenHands/pulls), with the evaluation process varying based on the type of change:
+We're generally happy to consider all pull requests with the evaluation process varying based on the type of change:
 
 #### For Small Improvements
 

Development.md

@@ -3,9 +3,9 @@ This guide is for people working on OpenHands and editing the source code.
 If you wish to contribute your changes, check out the [CONTRIBUTING.md](https://github.com/All-Hands-AI/OpenHands/blob/main/CONTRIBUTING.md) on how to clone and setup the project initially before moving on.
 Otherwise, you can clone the OpenHands project directly.
 
-## Start the server for development
+## Start the Server for Development
 ### 1. Requirements
-* Linux, Mac OS, or [WSL on Windows](https://learn.microsoft.com/en-us/windows/wsl/install)  [Ubuntu <= 22.04]
+* Linux, Mac OS, or [WSL on Windows](https://learn.microsoft.com/en-us/windows/wsl/install)  [Ubuntu >= 22.04]
 * [Docker](https://docs.docker.com/engine/install/) (For those on MacOS, make sure to allow the default Docker socket to be used from advanced settings!)
 * [Python](https://www.python.org/downloads/) = 3.12
 * [NodeJS](https://nodejs.org/en/download/package-manager) >= 20.x
@@ -58,7 +58,7 @@ See [our documentation](https://docs.all-hands.dev/modules/usage/llms) for recom
 
 ### 4. Running the application
 #### Option A: Run the Full Application
-Once the setup is complete, launching OpenHands is as simple as running a single command. This command starts both the backend and frontend servers seamlessly, allowing you to interact with OpenHands:
+Once the setup is complete, this command starts both the backend and frontend servers, allowing you to interact with OpenHands:
 ```bash
 make run
 ```
@@ -75,11 +75,11 @@ make run
     ```
 
 ### 6. LLM Debugging
-If you encounter any issues with the Language Model (LM) or you're simply curious, you can inspect the actual LLM prompts and responses. To do so, export DEBUG=1 in the environment and restart the backend.
-OpenHands will then log the prompts and responses in the logs/llm/CURRENT_DATE directory, allowing you to identify the causes.
+If you encounter any issues with the Language Model (LM) or you're simply curious, export DEBUG=1 in the environment and restart the backend.
+OpenHands will log the prompts and responses in the logs/llm/CURRENT_DATE directory, allowing you to identify the causes.
 
 ### 7. Help
-Need assistance or information on available targets and commands? The help command provides all the necessary guidance to ensure a smooth experience with OpenHands.
+Need help or info on available targets and commands? Use the help command for all the guidance you need with OpenHands.
 ```bash
 make help
  ```
@@ -93,14 +93,14 @@ poetry run pytest ./tests/unit/test_*.py
 ```
 
 ### 9. Add or update dependency
-1. Add your dependency in `pyproject.toml` or use `poetry add xxx`
-2. Update the poetry.lock file via `poetry lock --no-update`
+1. Add your dependency in `pyproject.toml` or use `poetry add xxx`.
+2. Update the poetry.lock file via `poetry lock --no-update`.
 
 ### 9. Use existing Docker image
 To reduce build time (e.g., if no changes were made to the client-runtime component), you can use an existing Docker container image by
 setting the SANDBOX_RUNTIME_CONTAINER_IMAGE environment variable to the desired Docker image.
 
-Example: `export SANDBOX_RUNTIME_CONTAINER_IMAGE=ghcr.io/all-hands-ai/runtime:0.18-nikolaik`
+Example: `export SANDBOX_RUNTIME_CONTAINER_IMAGE=ghcr.io/all-hands-ai/runtime:0.24-nikolaik`
 
 ## Develop inside Docker container
 
@@ -110,7 +110,7 @@ TL;DR
 make docker-dev
 ```
 
-See more details [here](./containers/dev/README.md)
+See more details [here](./containers/dev/README.md).
 
 If you are just interested in running `OpenHands` without installing all the required tools on your host.
 

ISSUE_TRIAGE.md

@@ -2,8 +2,8 @@
 These are the procedures and guidelines on how issues are triaged in this repo by the maintainers.
 
 ## General
-* Most issues must be tagged with **enhancement** or **bug**
-* Issues may be tagged with what it relates to (**backend**, **frontend**, **agent quality**, etc.)
+* Most issues must be tagged with **enhancement** or **bug**.
+* Issues may be tagged with what it relates to (**backend**, **frontend**, **agent quality**, etc.).
 
 ## Severity
 * **Low**: Minor issues or affecting single user.
@@ -11,10 +11,10 @@ These are the procedures and guidelines on how issues are triaged in this repo b
 * **Critical**: Affecting all users or potential security issues.
 
 ## Effort
-* Issues may be estimated with effort required (**small effort**, **medium effort**, **large effort**)
+* Issues may be estimated with effort required (**small effort**, **medium effort**, **large effort**).
 
 ## Difficulty
-* Issues with low implementation difficulty may be tagged with **good first issue**
+* Issues with low implementation difficulty may be tagged with **good first issue**.
 
 ## Not Enough Information
 * User is asked to provide more information (logs, how to reproduce, etc.) when the issue is not clear.

README.md

@@ -12,7 +12,7 @@
   <a href="https://codecov.io/github/All-Hands-AI/OpenHands?branch=main"><img alt="CodeCov" src="https://img.shields.io/codecov/c/github/All-Hands-AI/OpenHands?style=for-the-badge&color=blue"></a>
   <a href="https://github.com/All-Hands-AI/OpenHands/blob/main/LICENSE"><img src="https://img.shields.io/github/license/All-Hands-AI/OpenHands?style=for-the-badge&color=blue" alt="MIT License"></a>
   <br/>
-  <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2wkh4pklz-w~h_DVDtEe9H5kyQlcNxVw"><img src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge" alt="Join our Slack community"></a>
+  <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg"><img src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge" alt="Join our Slack community"></a>
   <a href="https://discord.gg/ESHStjSjD4"><img src="https://img.shields.io/badge/Discord-Join%20Us-purple?logo=discord&logoColor=white&style=for-the-badge" alt="Join our Discord community"></a>
   <a href="https://github.com/All-Hands-AI/OpenHands/blob/main/CREDITS.md"><img src="https://img.shields.io/badge/Project-Credits-blue?style=for-the-badge&color=FFE165&logo=github&logoColor=white" alt="Credits"></a>
   <br/>
@@ -39,21 +39,21 @@ Learn more at [docs.all-hands.dev](https://docs.all-hands.dev), or jump to the [
 ## ⚡ Quick Start
 
 The easiest way to run OpenHands is in Docker.
-See the [Installation](https://docs.all-hands.dev/modules/usage/installation) guide for
+See the [Running OpenHands](https://docs.all-hands.dev/modules/usage/installation) guide for
 system requirements and more information.
 
 ```bash
-docker pull docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik
 
 docker run -it --rm --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -e LOG_ALL_EVENTS=true \
     -v /var/run/docker.sock:/var/run/docker.sock \
     -v ~/.openhands-state:/.openhands-state \
     -p 3000:3000 \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app \
-    docker.all-hands.dev/all-hands-ai/openhands:0.18
+    docker.all-hands.dev/all-hands-ai/openhands:0.24
 ```
 
 You'll find OpenHands running at [http://localhost:3000](http://localhost:3000)!
@@ -69,7 +69,7 @@ run OpenHands in a scriptable [headless mode](https://docs.all-hands.dev/modules
 interact with it via a [friendly CLI](https://docs.all-hands.dev/modules/usage/how-to/cli-mode),
 or run it on tagged issues with [a github action](https://docs.all-hands.dev/modules/usage/how-to/github-action).
 
-Visit [Installation](https://docs.all-hands.dev/modules/usage/installation) for more information and setup instructions.
+Visit [Running OpenHands](https://docs.all-hands.dev/modules/usage/installation) for more information and setup instructions.
 
 > [!CAUTION]
 > OpenHands is meant to be run by a single user on their local workstation.
@@ -96,7 +96,7 @@ troubleshooting resources, and advanced configuration options.
 OpenHands is a community-driven project, and we welcome contributions from everyone. We do most of our communication
 through Slack, so this is the best place to start, but we also are happy to have you contact us on Discord or Github:
 
-- [Join our Slack workspace](https://join.slack.com/t/openhands-ai/shared_invite/zt-2wkh4pklz-w~h_DVDtEe9H5kyQlcNxVw) - Here we talk about research, architecture, and future development.
+- [Join our Slack workspace](https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg) - Here we talk about research, architecture, and future development.
 - [Join our Discord server](https://discord.gg/ESHStjSjD4) - This is a community-run server for general discussion, questions, and feedback.
 - [Read or post Github Issues](https://github.com/All-Hands-AI/OpenHands/issues) - Check out the issues we're working on, or add your own ideas.
 

build.sh

@@ -1,5 +1,4 @@
 #!/bin/bash
 set -e
 
-cp pyproject.toml poetry.lock openhands
 poetry build -v

config.template.toml

@@ -23,6 +23,9 @@ workspace_base = "./workspace"
 # Cache directory path
 #cache_dir = "/tmp/cache"
 
+# Reasoning effort for o1 models (low, medium, high, or not set)
+#reasoning_effort = "medium"
+
 # Debugging enabled
 #debug = false
 
@@ -34,7 +37,12 @@ workspace_base = "./workspace"
 
 # Path to store trajectories, can be a folder or a file
 # If it's a folder, the session id will be used as the file name
-#trajectories_path="./trajectories"
+#save_trajectory_path="./trajectories"
+
+# Path to replay a trajectory, must be a file path
+# If provided, trajectory will be loaded and replayed before the
+# agent responds to any user instruction
+#replay_trajectory_path = ""
 
 # File store path
 #file_store_path = "/tmp/file_store"
@@ -67,7 +75,7 @@ workspace_base = "./workspace"
 #run_as_openhands = true
 
 # Runtime environment
-#runtime = "eventstream"
+#runtime = "docker"
 
 # Name of the default agent
 #default_agent = "CodeActAgent"
@@ -96,7 +104,7 @@ workspace_base = "./workspace"
 #aws_secret_access_key = ""
 
 # API key to use (For Headless / CLI only -  In Web this is overridden by Session Init)
-api_key = "your-api-key"
+api_key = ""
 
 # API base URL (For Headless / CLI only -  In Web this is overridden by Session Init)
 #base_url = ""
@@ -187,7 +195,7 @@ model = "gpt-4o"
 #native_tool_calling = None
 
 [llm.gpt4o-mini]
-api_key = "your-api-key"
+api_key = ""
 model = "gpt-4o"
 
 
@@ -220,8 +228,8 @@ codeact_enable_jupyter = true
 # LLM config group to use
 #llm_config = 'your-llm-config-group'
 
-# Whether to use microagents at all
-#use_microagents = true
+# Whether to use prompt extension (e.g., microagent, repo/runtime info) at all
+#enable_prompt_extensions = true
 
 # List of microagents to disable
 #disabled_microagents = []

containers/dev/compose.yml

@@ -11,7 +11,7 @@ services:
       - BACKEND_HOST=${BACKEND_HOST:-"0.0.0.0"}
       - SANDBOX_API_HOSTNAME=host.docker.internal
       #
-      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-ghcr.io/all-hands-ai/runtime:0.18-nikolaik}
+      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-ghcr.io/all-hands-ai/runtime:0.24-nikolaik}
       - SANDBOX_USER_ID=${SANDBOX_USER_ID:-1234}
       - WORKSPACE_MOUNT_PATH=${WORKSPACE_BASE:-$PWD/workspace}
     ports:

dev_config/python/ruff.toml

@@ -24,3 +24,6 @@ inline-quotes = "single"
 
 [format]
 quote-style = "single"
+
+[lint.flake8-bugbear]
+extend-immutable-calls = ["Depends", "fastapi.Depends", "fastapi.params.Depends"]

docker-compose.yml

@@ -1,4 +1,4 @@
-#
+
 services:
   openhands:
     build:
@@ -7,8 +7,8 @@ services:
     image: openhands:latest
     container_name: openhands-app-${DATE:-}
     environment:
-      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-ghcr.io/all-hands-ai/runtime:0.18-nikolaik}
-      - SANDBOX_USER_ID=${SANDBOX_USER_ID:-1234}
+      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik}
+      #- SANDBOX_USER_ID=${SANDBOX_USER_ID:-1234} # enable this only if you want a specific non-root sandbox user but you will have to manually adjust permissions of openhands-state for this user
       - WORKSPACE_MOUNT_PATH=${WORKSPACE_BASE:-$PWD/workspace}
     ports:
       - "3000:3000"
@@ -16,6 +16,7 @@ services:
       - "host.docker.internal:host-gateway"
     volumes:
       - /var/run/docker.sock:/var/run/docker.sock
+      - ~/.openhands-state:/.openhands-state
       - ${WORKSPACE_BASE:-$PWD/workspace}:/opt/workspace_base
     pull_policy: build
     stdin_open: true

docs/i18n/fr/docusaurus-plugin-content-docs/current/usage/architecture/runtime.md

@@ -1,8 +1,8 @@
 
 
-# 📦 Runtime EventStream
+# 📦 Runtime Docker
 
-Le Runtime EventStream d'OpenHands est le composant principal qui permet l'exécution sécurisée et flexible des actions des agents d'IA.
+Le Runtime Docker d'OpenHands est le composant principal qui permet l'exécution sécurisée et flexible des actions des agents d'IA.
 Il crée un environnement en bac à sable (sandbox) en utilisant Docker, où du code arbitraire peut être exécuté en toute sécurité sans risquer le système hôte.
 
 ## Pourquoi avons-nous besoin d'un runtime en bac à sable ?

docs/i18n/fr/docusaurus-plugin-content-docs/current/usage/configuration-options.md

@@ -1,5 +1,3 @@
-
-
 # Options de configuration
 
 Ce guide détaille toutes les options de configuration disponibles pour OpenHands, vous aidant à personnaliser son comportement et à l'intégrer avec d'autres services.
@@ -94,7 +92,7 @@ Les options de configuration de base sont définies dans la section `[core]` du
   - Description : Désactiver la couleur dans la sortie du terminal
 
 **Trajectoires**
-- `trajectories_path`
+- `save_trajectory_path`
   - Type : `str`
   - Valeur par défaut : `"./trajectories"`
   - Description : Chemin pour stocker les trajectoires (peut être un dossier ou un fichier). Si c'est un dossier, les trajectoires seront enregistrées dans un fichier nommé avec l'ID de session et l'extension .json, dans ce dossier.
@@ -165,7 +163,7 @@ Les options de configuration de base sont définies dans la section `[core]` du
 
 - `runtime`
   - Type : `str`
-  - Valeur par défaut : `"eventstream"`
+  - Valeur par défaut : `"docker"`
   - Description : Environnement d'exécution
 
 - `default_agent`
@@ -184,6 +182,10 @@ Les options de configuration LLM (Large Language Model) sont définies dans la s
 
 Pour les utiliser avec la commande docker, passez `-e LLM_<option>`. Exemple : `-e LLM_NUM_RETRIES`.
 
+:::note
+Pour les configurations de développement, vous pouvez également définir des configurations LLM personnalisées. Voir [Configurations LLM personnalisées](./llms/custom-llm-configs) pour plus de détails.
+:::
+
 **Informations d'identification AWS**
 - `aws_access_key_id`
   - Type : `str`
@@ -368,4 +370,26 @@ Les options de configuration de l'agent sont définies dans les sections `[agent
 - `codeact_enable_llm_editor`
   - Type : `bool`
   - Valeur par défaut : `false`
-  - Description : Si l'éditeur LLM est activé dans l'espace d'action (foncti
+  - Description : Si l'éditeur LLM est activé dans l'espace d'action (fonctionne uniquement avec l'appel de fonction)
+
+**Utilisation du micro-agent**
+- `enable_prompt_extensions`
+  - Type : `bool`
+  - Valeur par défaut : `true`
+  - Description : Indique si l'utilisation des micro-agents est activée ou non
+
+- `disabled_microagents`
+  - Type : `list of str`
+  - Valeur par défaut : `None`
+  - Description : Liste des micro-agents à désactiver
+
+### Exécution
+- `timeout`
+  - Type : `int`
+  - Valeur par défaut : `120`
+  - Description : Délai d'expiration du bac à sable, en secondes
+
+- `user_id`
+  - Type : `int`
+  - Valeur par défaut : `1000`
+  - Description : ID de l'utilisateur du bac à sable

docs/i18n/fr/docusaurus-plugin-content-docs/current/usage/custom_sandbox_guide.md

@@ -95,7 +95,3 @@ sandbox_user_id="1001"
 ### Erreurs de port d'utilisation
 
 Si vous voyez un message d'erreur indiquant que le port est utilisé ou indisponible, essayez de supprimer toutes les containers docker en cours d'exécution (exécutez `docker ps` et `docker rm` des containers concernés) puis ré-exécutez ```make run```
-
-## Discuter
-
-Pour d'autres problèmes ou questions rejoignez le [Slack](https://join.slack.com/t/openhands-ai/shared_invite/zt-2wkh4pklz-w~h_DVDtEe9H5kyQlcNxVw) ou le [Discord](https://discord.gg/ESHStjSjD4) et demandez!

docs/i18n/fr/docusaurus-plugin-content-docs/current/usage/how-to/cli-mode.md

@@ -52,7 +52,7 @@ LLM_API_KEY="sk_test_12345"
 ```bash
 docker run -it \
     --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -e SANDBOX_USER_ID=$(id -u) \
     -e WORKSPACE_MOUNT_PATH=$WORKSPACE_BASE \
     -e LLM_API_KEY=$LLM_API_KEY \
@@ -61,7 +61,7 @@ docker run -it \
     -v /var/run/docker.sock:/var/run/docker.sock \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app-$(date +%Y%m%d%H%M%S) \
-    docker.all-hands.dev/all-hands-ai/openhands:0.18 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.24 \
     python -m openhands.core.cli
 ```
 

docs/i18n/fr/docusaurus-plugin-content-docs/current/usage/how-to/evaluation-harness.md

@@ -114,7 +114,7 @@ Pour créer un workflow d'évaluation pour votre benchmark, suivez ces étapes :
    def get_config(instance: pd.Series, metadata: EvalMetadata) -> AppConfig:
        config = AppConfig(
            default_agent=metadata.agent_class,
-           runtime='eventstream',
+           runtime='docker',
            max_iterations=metadata.max_iterations,
            sandbox=SandboxConfig(
                base_container_image='your_container_image',

docs/i18n/fr/docusaurus-plugin-content-docs/current/usage/how-to/headless-mode.md

@@ -46,7 +46,7 @@ LLM_API_KEY="sk_test_12345"
 ```bash
 docker run -it \
     --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -e SANDBOX_USER_ID=$(id -u) \
     -e WORKSPACE_MOUNT_PATH=$WORKSPACE_BASE \
     -e LLM_API_KEY=$LLM_API_KEY \
@@ -56,6 +56,6 @@ docker run -it \
     -v /var/run/docker.sock:/var/run/docker.sock \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app-$(date +%Y%m%d%H%M%S) \
-    docker.all-hands.dev/all-hands-ai/openhands:0.18 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.24 \
     python -m openhands.core.main -t "write a bash script that prints hi" --no-auto-continue
 ```

docs/i18n/fr/docusaurus-plugin-content-docs/current/usage/installation.mdx

@@ -13,16 +13,16 @@
 La façon la plus simple d'exécuter OpenHands est avec Docker.
 
 ```bash
-docker pull docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik
 
 docker run -it --rm --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -e LOG_ALL_EVENTS=true \
     -v /var/run/docker.sock:/var/run/docker.sock \
     -p 3000:3000 \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app \
-    docker.all-hands.dev/all-hands-ai/openhands:0.18
+    docker.all-hands.dev/all-hands-ai/openhands:0.24
 ```
 
 Vous pouvez également exécuter OpenHands en mode [headless scriptable](https://docs.all-hands.dev/modules/usage/how-to/headless-mode), en tant que [CLI interactive](https://docs.all-hands.dev/modules/usage/how-to/cli-mode), ou en utilisant l'[Action GitHub OpenHands](https://docs.all-hands.dev/modules/usage/how-to/github-action).

docs/i18n/fr/docusaurus-plugin-content-docs/current/usage/intro.mdx

@@ -42,7 +42,7 @@ Explorez le code source d'OpenHands sur [GitHub](https://github.com/All-Hands-AI
   />
 </a>
 <br></br>
-<a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2wkh4pklz-w~h_DVDtEe9H5kyQlcNxVw">
+<a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg">
   <img
     src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge"
     alt="Join our Slack community"

docs/i18n/fr/docusaurus-plugin-content-docs/current/usage/llms/custom-llm-configs.md

@@ -0,0 +1,106 @@
+# Configurations LLM personnalisées
+
+OpenHands permet de définir plusieurs configurations LLM nommées dans votre fichier `config.toml`. Cette fonctionnalité vous permet d'utiliser différentes configurations LLM pour différents usages, comme utiliser un modèle moins coûteux pour les tâches qui ne nécessitent pas de réponses de haute qualité, ou utiliser différents modèles avec différents paramètres pour des agents spécifiques.
+
+## Comment ça fonctionne
+
+Les configurations LLM nommées sont définies dans le fichier `config.toml` en utilisant des sections qui commencent par `llm.`. Par exemple :
+
+```toml
+# Configuration LLM par défaut
+[llm]
+model = "gpt-4"
+api_key = "votre-clé-api"
+temperature = 0.0
+
+# Configuration LLM personnalisée pour un modèle moins coûteux
+[llm.gpt3]
+model = "gpt-3.5-turbo"
+api_key = "votre-clé-api"
+temperature = 0.2
+
+# Une autre configuration personnalisée avec des paramètres différents
+[llm.haute-creativite]
+model = "gpt-4"
+api_key = "votre-clé-api"
+temperature = 0.8
+top_p = 0.9
+```
+
+Chaque configuration nommée hérite de tous les paramètres de la section `[llm]` par défaut et peut remplacer n'importe lequel de ces paramètres. Vous pouvez définir autant de configurations personnalisées que nécessaire.
+
+## Utilisation des configurations personnalisées
+
+### Avec les agents
+
+Vous pouvez spécifier quelle configuration LLM un agent doit utiliser en définissant le paramètre `llm_config` dans la section de configuration de l'agent :
+
+```toml
+[agent.RepoExplorerAgent]
+# Utiliser la configuration GPT-3 moins coûteuse pour cet agent
+llm_config = 'gpt3'
+
+[agent.CodeWriterAgent]
+# Utiliser la configuration haute créativité pour cet agent
+llm_config = 'haute-creativite'
+```
+
+### Options de configuration
+
+Chaque configuration LLM nommée prend en charge toutes les mêmes options que la configuration LLM par défaut. Celles-ci incluent :
+
+- Sélection du modèle (`model`)
+- Configuration de l'API (`api_key`, `base_url`, etc.)
+- Paramètres du modèle (`temperature`, `top_p`, etc.)
+- Paramètres de nouvelle tentative (`num_retries`, `retry_multiplier`, etc.)
+- Limites de jetons (`max_input_tokens`, `max_output_tokens`)
+- Et toutes les autres options de configuration LLM
+
+Pour une liste complète des options disponibles, consultez la section Configuration LLM dans la documentation des [Options de configuration](../configuration-options).
+
+## Cas d'utilisation
+
+Les configurations LLM personnalisées sont particulièrement utiles dans plusieurs scénarios :
+
+- **Optimisation des coûts** : Utiliser des modèles moins coûteux pour les tâches qui ne nécessitent pas de réponses de haute qualité, comme l'exploration de dépôt ou les opérations simples sur les fichiers.
+- **Réglage spécifique aux tâches** : Configurer différentes valeurs de température et de top_p pour les tâches qui nécessitent différents niveaux de créativité ou de déterminisme.
+- **Différents fournisseurs** : Utiliser différents fournisseurs LLM ou points d'accès API pour différentes tâches.
+- **Tests et développement** : Basculer facilement entre différentes configurations de modèles pendant le développement et les tests.
+
+## Exemple : Optimisation des coûts
+
+Un exemple pratique d'utilisation des configurations LLM personnalisées pour optimiser les coûts :
+
+```toml
+# Configuration par défaut utilisant GPT-4 pour des réponses de haute qualité
+[llm]
+model = "gpt-4"
+api_key = "votre-clé-api"
+temperature = 0.0
+
+# Configuration moins coûteuse pour l'exploration de dépôt
+[llm.repo-explorer]
+model = "gpt-3.5-turbo"
+temperature = 0.2
+
+# Configuration pour la génération de code
+[llm.code-gen]
+model = "gpt-4"
+temperature = 0.0
+max_output_tokens = 2000
+
+[agent.RepoExplorerAgent]
+llm_config = 'repo-explorer'
+
+[agent.CodeWriterAgent]
+llm_config = 'code-gen'
+```
+
+Dans cet exemple :
+- L'exploration de dépôt utilise un modèle moins coûteux car il s'agit principalement de comprendre et de naviguer dans le code
+- La génération de code utilise GPT-4 avec une limite de jetons plus élevée pour générer des blocs de code plus importants
+- La configuration par défaut reste disponible pour les autres tâches
+
+:::note
+Les configurations LLM personnalisées ne sont disponibles que lors de l'utilisation d'OpenHands en mode développement, via `main.py` ou `cli.py`. Lors de l'exécution via `docker run`, veuillez utiliser les options de configuration standard.
+:::

docs/i18n/fr/docusaurus-plugin-content-docs/current/usage/runtimes.md

@@ -13,7 +13,7 @@ C'est le Runtime par défaut qui est utilisé lorsque vous démarrez OpenHands.
 
 ```
 docker run # ...
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -v /var/run/docker.sock:/var/run/docker.sock \
     # ...
 ```

docs/i18n/zh-Hans/docusaurus-plugin-content-docs/current/usage/architecture/runtime.md

@@ -1,8 +1,8 @@
 以下是翻译后的内容:
 
-# 📦 EventStream 运行时
+# 📦 Docker 运行时
 
-OpenHands EventStream 运行时是实现 AI 代理操作安全灵活执行的核心组件。
+OpenHands Docker 运行时是实现 AI 代理操作安全灵活执行的核心组件。
 它使用 Docker 创建一个沙盒环境,可以安全地运行任意代码而不会危及主机系统。
 
 ## 为什么我们需要沙盒运行时?

docs/i18n/zh-Hans/docusaurus-plugin-content-docs/current/usage/configuration-options.md

@@ -91,7 +91,7 @@
   - 描述: 禁用终端输出中的颜色
 
 **轨迹**
-- `trajectories_path`
+- `save_trajectory_path`
   - 类型: `str`
   - 默认值: `"./trajectories"`
   - 描述: 存储轨迹的路径(可以是文件夹或文件)。如果是文件夹,轨迹将保存在该文件夹中以会话 ID 命名的 .json 文件中。
@@ -162,7 +162,7 @@
 
 - `runtime`
   - 类型: `str`
-  - 默认值: `"eventstream"`
+  - 默认值: `"docker"`
   - 描述: 运行时环境
 
 - `default_agent`
@@ -373,7 +373,7 @@ Agent 配置选项在 `config.toml` 文件的 `[agent]` 和 `[agent.<agent_name>
   - 描述: 是否在 action space 中启用 Jupyter
 
 **Microagent 使用**
-- `use_microagents`
+- `enable_prompt_extensions`
   - 类型: `bool`
   - 默认值: `true`
   - 描述: 是否使用 microagents

docs/i18n/zh-Hans/docusaurus-plugin-content-docs/current/usage/custom_sandbox_guide.md

@@ -96,7 +96,3 @@ sandbox_user_id="1001"
 ### 端口使用错误
 
 如果您遇到端口被占用或不可用的错误提示，可以尝试先用`docker ps`命令列出所有运行中的 Docker 容器，然后使用`docker rm`命令删除相关容器，最后再重新执行```make run```命令。
-
-## 讨论
-
-对于其他问题或疑问，请加入 [Slack](https://join.slack.com/t/openhands-ai/shared_invite/zt-2wkh4pklz-w~h_DVDtEe9H5kyQlcNxVw) 或 [Discord](https://discord.gg/ESHStjSjD4) 提问！

docs/i18n/zh-Hans/docusaurus-plugin-content-docs/current/usage/how-to/cli-mode.md

@@ -50,7 +50,7 @@ LLM_API_KEY="sk_test_12345"
 ```bash
 docker run -it \
     --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -e SANDBOX_USER_ID=$(id -u) \
     -e WORKSPACE_MOUNT_PATH=$WORKSPACE_BASE \
     -e LLM_API_KEY=$LLM_API_KEY \
@@ -59,7 +59,7 @@ docker run -it \
     -v /var/run/docker.sock:/var/run/docker.sock \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app-$(date +%Y%m%d%H%M%S) \
-    docker.all-hands.dev/all-hands-ai/openhands:0.18 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.24 \
     python -m openhands.core.cli
 ```
 

docs/i18n/zh-Hans/docusaurus-plugin-content-docs/current/usage/how-to/evaluation-harness.md

@@ -112,7 +112,7 @@ OpenHands 的主要入口点在 `openhands/core/main.py` 中。以下是它的
    def get_config(instance: pd.Series, metadata: EvalMetadata) -> AppConfig:
        config = AppConfig(
            default_agent=metadata.agent_class,
-           runtime='eventstream',
+           runtime='docker',
            max_iterations=metadata.max_iterations,
            sandbox=SandboxConfig(
                base_container_image='your_container_image',

docs/i18n/zh-Hans/docusaurus-plugin-content-docs/current/usage/how-to/headless-mode.md

@@ -47,7 +47,7 @@ LLM_API_KEY="sk_test_12345"
 ```bash
 docker run -it \
     --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -e SANDBOX_USER_ID=$(id -u) \
     -e WORKSPACE_MOUNT_PATH=$WORKSPACE_BASE \
     -e LLM_API_KEY=$LLM_API_KEY \
@@ -57,6 +57,6 @@ docker run -it \
     -v /var/run/docker.sock:/var/run/docker.sock \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app-$(date +%Y%m%d%H%M%S) \
-    docker.all-hands.dev/all-hands-ai/openhands:0.18 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.24 \
     python -m openhands.core.main -t "write a bash script that prints hi" --no-auto-continue
 ```

docs/i18n/zh-Hans/docusaurus-plugin-content-docs/current/usage/installation.mdx

@@ -11,16 +11,16 @@
 在 Docker 中运行 OpenHands 是最简单的方式。
 
 ```bash
-docker pull docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik
 
 docker run -it --rm --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -e LOG_ALL_EVENTS=true \
     -v /var/run/docker.sock:/var/run/docker.sock \
     -p 3000:3000 \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app \
-    docker.all-hands.dev/all-hands-ai/openhands:0.18
+    docker.all-hands.dev/all-hands-ai/openhands:0.24
 ```
 
 你也可以在可脚本化的[无头模式](https://docs.all-hands.dev/modules/usage/how-to/headless-mode)下运行 OpenHands，作为[交互式 CLI](https://docs.all-hands.dev/modules/usage/how-to/cli-mode)，或使用 [OpenHands GitHub Action](https://docs.all-hands.dev/modules/usage/how-to/github-action)。

docs/i18n/zh-Hans/docusaurus-plugin-content-docs/current/usage/intro.mdx

@@ -42,7 +42,7 @@ OpenHands 是一个**自主 AI 软件工程师**，能够执行复杂的工程
   />
 </a>
 <br></br>
-<a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2wkh4pklz-w~h_DVDtEe9H5kyQlcNxVw">
+<a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg">
   <img
     src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge"
     alt="Join our Slack community"

docs/i18n/zh-Hans/docusaurus-plugin-content-docs/current/usage/runtimes.md

@@ -11,7 +11,7 @@
 
 ```
 docker run # ...
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -v /var/run/docker.sock:/var/run/docker.sock \
     # ...
 ```

docs/modules/usage/architecture/runtime.md

@@ -1,6 +1,6 @@
-# 📦 EventStream Runtime
+# 📦 Docker Runtime
 
-The OpenHands EventStream Runtime is the core component that enables secure and flexible execution of AI agent's action.
+The OpenHands Docker Runtime is the core component that enables secure and flexible execution of AI agent's action.
 It creates a sandboxed environment using Docker, where arbitrary code can be run safely without risking the host system.
 
 ## Why do we need a sandboxed runtime?

docs/modules/usage/configuration-options.md

@@ -7,53 +7,11 @@ If you are running in [GUI Mode](https://docs.all-hands.dev/modules/usage/how-to
 take precedence.
 :::
 
----
-
-# Table of Contents
-
-- [Core Configuration](#core-configuration)
-   - [API Keys](#api-keys)
-   - [Workspace](#workspace)
-   - [Debugging and Logging](#debugging-and-logging)
-   - [Session Management](#session-management)
-   - [Trajectories](#trajectories)
-   - [File Store](#file-store)
-   - [Task Management](#task-management)
-   - [Sandbox Configuration](#sandbox-configuration)
-   - [Miscellaneous](#miscellaneous)
-- [LLM Configuration](#llm-configuration)
-   - [AWS Credentials](#aws-credentials)
-   - [API Configuration](#api-configuration)
-   - [Custom LLM Provider](#custom-llm-provider)
-   - [Embeddings](#embeddings)
-   - [Message Handling](#message-handling)
-   - [Model Selection](#model-selection)
-   - [Retrying](#retrying)
-   - [Advanced Options](#advanced-options)
-- [Agent Configuration](#agent-configuration)
-   - [Microagent Configuration](#microagent-configuration)
-   - [Memory Configuration](#memory-configuration)
-   - [LLM Configuration](#llm-configuration-2)
-   - [ActionSpace Configuration](#actionspace-configuration)
-   - [Microagent Usage](#microagent-usage)
-- [Sandbox Configuration](#sandbox-configuration)
-   - [Execution](#execution)
-   - [Container Image](#container-image)
-   - [Networking](#networking)
-   - [Linting and Plugins](#linting-and-plugins)
-   - [Dependencies and Environment](#dependencies-and-environment)
-   - [Evaluation](#evaluation)
-- [Security Configuration](#security-configuration)
-   - [Confirmation Mode](#confirmation-mode)
-   - [Security Analyzer](#security-analyzer)
-
----
-
 ## Core Configuration
 
 The core configuration options are defined in the `[core]` section of the `config.toml` file.
 
-**API Keys**
+### API Keys
 - `e2b_api_key`
   - Type: `str`
   - Default: `""`
@@ -69,7 +27,7 @@ The core configuration options are defined in the `[core]` section of the `confi
   - Default: `""`
   - Description: API token secret for Modal
 
-**Workspace**
+### Workspace
 - `workspace_base`
   - Type: `str`
   - Default: `"./workspace"`
@@ -80,7 +38,7 @@ The core configuration options are defined in the `[core]` section of the `confi
   - Default: `"/tmp/cache"`
   - Description: Cache directory path
 
-**Debugging and Logging**
+### Debugging and Logging
 - `debug`
   - Type: `bool`
   - Default: `false`
@@ -91,13 +49,18 @@ The core configuration options are defined in the `[core]` section of the `confi
   - Default: `false`
   - Description: Disable color in terminal output
 
-**Trajectories**
-- `trajectories_path`
+### Trajectories
+- `save_trajectory_path`
   - Type: `str`
   - Default: `"./trajectories"`
   - Description: Path to store trajectories (can be a folder or a file). If it's a folder, the trajectories will be saved in a file named with the session id name and .json extension, in that folder.
 
-**File Store**
+- `replay_trajectory_path`
+  - Type: `str`
+  - Default: `""`
+  - Description: Path to load a trajectory and replay. If given, must be a path to the trajectory file in JSON format. The actions in the trajectory file would be replayed first before any user instruction is executed.
+
+### File Store
 - `file_store_path`
   - Type: `str`
   - Default: `"/tmp/file_store"`
@@ -128,7 +91,7 @@ The core configuration options are defined in the `[core]` section of the `confi
   - Default: `[".*"]`
   - Description: List of allowed file extensions for uploads
 
-**Task Management**
+### Task Management
 - `max_budget_per_task`
   - Type: `float`
   - Default: `0.0`
@@ -139,7 +102,7 @@ The core configuration options are defined in the `[core]` section of the `confi
   - Default: `100`
   - Description: Maximum number of iterations
 
-**Sandbox Configuration**
+### Sandbox Configuration
 - `workspace_mount_path_in_sandbox`
   - Type: `str`
   - Default: `"/workspace"`
@@ -155,7 +118,7 @@ The core configuration options are defined in the `[core]` section of the `confi
   - Default: `""`
   - Description: Path to rewrite the workspace mount path to. You can usually ignore this, it refers to special cases of running inside another container.
 
-**Miscellaneous**
+### Miscellaneous
 - `run_as_openhands`
   - Type: `bool`
   - Default: `true`
@@ -163,7 +126,7 @@ The core configuration options are defined in the `[core]` section of the `confi
 
 - `runtime`
   - Type: `str`
-  - Default: `"eventstream"`
+  - Default: `"docker"`
   - Description: Runtime environment
 
 - `default_agent`
@@ -182,6 +145,10 @@ The LLM (Large Language Model) configuration options are defined in the `[llm]`
 
 To use these with the docker command, pass in `-e LLM_<option>`. Example: `-e LLM_NUM_RETRIES`.
 
+:::note
+For development setups, you can also define custom named LLM configurations. See [Custom LLM Configurations](./llms/custom-llm-configs) for details.
+:::
+
 **AWS Credentials**
 - `aws_access_key_id`
   - Type: `str`
@@ -198,7 +165,7 @@ To use these with the docker command, pass in `-e LLM_<option>`. Example: `-e LL
   - Default: `""`
   - Description: AWS secret access key
 
-**API Configuration**
+### API Configuration
 - `api_key`
   - Type: `str`
   - Default: `None`
@@ -224,13 +191,13 @@ To use these with the docker command, pass in `-e LLM_<option>`. Example: `-e LL
   - Default: `0.0`
   - Description: Cost per output token
 
-**Custom LLM Provider**
+### Custom LLM Provider
 - `custom_llm_provider`
   - Type: `str`
   - Default: `""`
   - Description: Custom LLM provider
 
-**Embeddings**
+### Embeddings
 - `embedding_base_url`
   - Type: `str`
   - Default: `""`
@@ -246,7 +213,7 @@ To use these with the docker command, pass in `-e LLM_<option>`. Example: `-e LL
   - Default: `"local"`
   - Description: Embedding model to use
 
-**Message Handling**
+### Message Handling
 - `max_message_chars`
   - Type: `int`
   - Default: `30000`
@@ -262,13 +229,13 @@ To use these with the docker command, pass in `-e LLM_<option>`. Example: `-e LL
   - Default: `0`
   - Description: Maximum number of output tokens
 
-**Model Selection**
+### Model Selection
 - `model`
   - Type: `str`
   - Default: `"claude-3-5-sonnet-20241022"`
   - Description: Model to use
 
-**Retrying**
+### Retrying
 - `num_retries`
   - Type: `int`
   - Default: `8`
@@ -289,7 +256,7 @@ To use these with the docker command, pass in `-e LLM_<option>`. Example: `-e LL
   - Default: `2.0`
   - Description: Multiplier for exponential backoff calculation
 
-**Advanced Options**
+### Advanced Options
 - `drop_params`
   - Type: `bool`
   - Default: `false`
@@ -329,13 +296,13 @@ To use these with the docker command, pass in `-e LLM_<option>`. Example: `-e LL
 
 The agent configuration options are defined in the `[agent]` and `[agent.<agent_name>]` sections of the `config.toml` file.
 
-**Microagent Configuration**
+### Microagent Configuration
 - `micro_agent_name`
   - Type: `str`
   - Default: `""`
   - Description: Name of the micro agent to use for this agent
 
-**Memory Configuration**
+### Memory Configuration
 - `memory_enabled`
   - Type: `bool`
   - Default: `false`
@@ -346,13 +313,13 @@ The agent configuration options are defined in the `[agent]` and `[agent.<agent_
   - Default: `3`
   - Description: The maximum number of threads indexing at the same time for embeddings
 
-**LLM Configuration**
+### LLM Configuration
 - `llm_config`
   - Type: `str`
   - Default: `'your-llm-config-group'`
   - Description: The name of the LLM config to use
 
-**ActionSpace Configuration**
+### ActionSpace Configuration
 - `function_calling`
   - Type: `bool`
   - Default: `true`
@@ -373,8 +340,8 @@ The agent configuration options are defined in the `[agent]` and `[agent.<agent_
   - Default: `false`
   - Description: Whether Jupyter is enabled in the action space
 
-**Microagent Usage**
-- `use_microagents`
+### Microagent Usage
+- `enable_prompt_extensions`
   - Type: `bool`
   - Default: `true`
   - Description: Whether to use microagents at all
@@ -390,7 +357,7 @@ The sandbox configuration options are defined in the `[sandbox]` section of the
 
 To use these with the docker command, pass in `-e SANDBOX_<option>`. Example: `-e SANDBOX_TIMEOUT`.
 
-**Execution**
+### Execution
 - `timeout`
   - Type: `int`
   - Default: `120`
@@ -401,19 +368,19 @@ To use these with the docker command, pass in `-e SANDBOX_<option>`. Example: `-
   - Default: `1000`
   - Description: Sandbox user ID
 
-**Container Image**
+### Container Image
 - `base_container_image`
   - Type: `str`
   - Default: `"nikolaik/python-nodejs:python3.12-nodejs22"`
   - Description: Container image to use for the sandbox
 
-**Networking**
+### Networking
 - `use_host_network`
   - Type: `bool`
   - Default: `false`
   - Description: Use host network
 
-**Linting and Plugins**
+### Linting and Plugins
 - `enable_auto_lint`
   - Type: `bool`
   - Default: `false`
@@ -424,7 +391,7 @@ To use these with the docker command, pass in `-e SANDBOX_<option>`. Example: `-
   - Default: `true`
   - Description: Whether to initialize plugins
 
-**Dependencies and Environment**
+### Dependencies and Environment
 - `runtime_extra_deps`
   - Type: `str`
   - Default: `""`
@@ -435,7 +402,7 @@ To use these with the docker command, pass in `-e SANDBOX_<option>`. Example: `-
   - Default: `{}`
   - Description: Environment variables to set at the launch of the runtime
 
-**Evaluation**
+### Evaluation
 - `browsergym_eval_env`
   - Type: `str`
   - Default: `""`
@@ -447,13 +414,13 @@ The security configuration options are defined in the `[security]` section of th
 
 To use these with the docker command, pass in `-e SECURITY_<option>`. Example: `-e SECURITY_CONFIRMATION_MODE`.
 
-**Confirmation Mode**
+### Confirmation Mode
 - `confirmation_mode`
   - Type: `bool`
   - Default: `false`
   - Description: Enable confirmation mode
 
-**Security Analyzer**
+### Security Analyzer
 - `security_analyzer`
   - Type: `str`
   - Default: `""`

docs/modules/usage/getting-started.mdx

@@ -1,6 +1,6 @@
 # Getting Started with OpenHands
 
-So you've [installed OpenHands](./installation) and have
+So you've [run OpenHands](./installation) and have
 [set up your LLM](./installation#setup). Now what?
 
 OpenHands can help you tackle a wide variety of engineering tasks. But the technology

docs/modules/usage/how-to/cli-mode.md

@@ -35,7 +35,7 @@ To run OpenHands in CLI mode with Docker:
 ```bash
 docker run -it \
     --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -e SANDBOX_USER_ID=$(id -u) \
     -e WORKSPACE_MOUNT_PATH=$WORKSPACE_BASE \
     -e LLM_API_KEY=$LLM_API_KEY \
@@ -45,7 +45,7 @@ docker run -it \
     -v ~/.openhands-state:/.openhands-state \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app-$(date +%Y%m%d%H%M%S) \
-    docker.all-hands.dev/all-hands-ai/openhands:0.18 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.24 \
     python -m openhands.core.cli
 ```
 
@@ -58,7 +58,7 @@ Here are some examples of CLI commands and their expected outputs:
 ### Example 1: Simple Task
 
 ```bash
-How can I help? >> Write a Python script that prints "Hello, World!"
+>> Write a Python script that prints "Hello, World!"
 ```
 
 Expected Output:
@@ -72,7 +72,7 @@ Expected Output:
 ### Example 2: Bash Command
 
 ```bash
-How can I help? >> Create a directory named "test_dir"
+>> Create a directory named "test_dir"
 ```
 
 Expected Output:
@@ -86,7 +86,7 @@ Expected Output:
 ### Example 3: Error Handling
 
 ```bash
-How can I help? >> Delete a non-existent file
+>> Delete a non-existent file
 ```
 
 Expected Output:

docs/modules/usage/how-to/custom-sandbox-guide.md

@@ -18,15 +18,21 @@ If you choose the first option, you can skip the `Create Your Docker Image` sect
 
 To create a custom Docker image, it must be Debian based.
 
-For example, if you want OpenHands to have `ruby` installed, create a `Dockerfile` with the following content:
+For example, if you want OpenHands to have `ruby` installed, you could create a `Dockerfile` with the following content:
 
 ```dockerfile
-FROM debian:latest
+FROM nikolaik/python-nodejs:python3.12-nodejs22
 
 # Install required packages
 RUN apt-get update && apt-get install -y ruby
 ```
 
+Or you could use a Ruby-specific base image:
+
+```dockerfile
+FROM ruby:latest
+```
+
 Save this file in a folder. Then, build your Docker image (e.g., named custom-image) by navigating to the folder in
 the terminal and running::
 ```bash
@@ -35,8 +41,16 @@ docker build -t custom-image .
 
 This will produce a new image called `custom-image`, which will be available in Docker.
 
-> Note that in the configuration described in this document, OpenHands will run as user "openhands" inside the
-> sandbox and thus all packages installed via the docker file should be available to all users on the system, not just root.
+## Using the Docker Command
+
+When running OpenHands using [the docker command](/modules/usage/installation#start-the-app), replace
+`-e SANDBOX_RUNTIME_CONTAINER_IMAGE=...` with `-e SANDBOX_BASE_CONTAINER_IMAGE=<custom image name>`:
+
+```commandline
+docker run -it --rm --pull=always \
+    -e SANDBOX_BASE_CONTAINER_IMAGE=custom-image \
+    ...
+```
 
 ## Using the Development Workflow
 
@@ -55,6 +69,28 @@ This can be an image you’ve already pulled or one you’ve built:
 sandbox_base_container_image="custom-image"
 ```
 
+### Additional Configuration Options
+
+The `config.toml` file supports several other options for customizing your sandbox:
+
+```toml
+[core]
+# Install additional dependencies when the runtime is built
+# Can contain any valid shell commands
+# If you need the path to the Python interpreter in any of these commands, you can use the $OH_INTERPRETER_PATH variable
+runtime_extra_deps = """
+pip install numpy pandas
+apt-get update && apt-get install -y ffmpeg
+"""
+
+# Set environment variables for the runtime
+# Useful for configuration that needs to be available at runtime
+runtime_startup_env_vars = { DATABASE_URL = "postgresql://user:pass@localhost/db" }
+
+# Specify platform for multi-architecture builds (e.g., "linux/amd64" or "linux/arm64")
+platform = "linux/amd64"
+```
+
 ### Run
 
 Run OpenHands by running ```make run``` in the top level directory.

docs/modules/usage/how-to/evaluation-harness.md

@@ -112,7 +112,7 @@ To create an evaluation workflow for your benchmark, follow these steps:
    def get_config(instance: pd.Series, metadata: EvalMetadata) -> AppConfig:
        config = AppConfig(
            default_agent=metadata.agent_class,
-           runtime='eventstream',
+           runtime='docker',
            max_iterations=metadata.max_iterations,
            sandbox=SandboxConfig(
                base_container_image='your_container_image',

docs/modules/usage/how-to/github-action.md

@@ -42,9 +42,10 @@ You can provide custom directions for OpenHands by following the [README for the
 Github resolver will automatically check for valid [repository secrets](https://docs.github.com/en/actions/security-for-github-actions/security-guides/using-secrets-in-github-actions?tool=webui#creating-secrets-for-a-repository) or [repository variables](https://docs.github.com/en/actions/writing-workflows/choosing-what-your-workflow-does/store-information-in-variables#creating-configuration-variables-for-a-repository) to customize its behavior.
 The customization options you can set are:
 
-| **Attribute name**               | **Type** | **Purpose**                                                                                                 | **Example**                                          |
-|----------------------------------| -------- |-------------------------------------------------------------------------------------------------------------|------------------------------------------------------|
-| `LLM_MODEL`                      | Variable | Set the LLM to use with OpenHands                                                                           | `LLM_MODEL="anthropic/claude-3-5-sonnet-20241022"`   |
-| `OPENHANDS_MAX_ITER`             | Variable | Set max limit for agent iterations                                                                          | `OPENHANDS_MAX_ITER=10`                              |
-| `OPENHANDS_MACRO`                | Variable | Customize default macro for invoking the resolver                                                           | `OPENHANDS_MACRO=@resolveit`                         |
-| `OPENHANDS_BASE_CONTAINER_IMAGE` | Variable | Custom Sandbox ([learn more](https://docs.all-hands.dev/modules/usage/how-to/custom-sandbox-guide))         | `OPENHANDS_BASE_CONTAINER_IMAGE="custom_image"`      |
+| **Attribute name**               | **Type** | **Purpose**                                                                                         | **Example**                                        |
+| -------------------------------- | -------- | --------------------------------------------------------------------------------------------------- | -------------------------------------------------- |
+| `LLM_MODEL`                      | Variable | Set the LLM to use with OpenHands                                                                   | `LLM_MODEL="anthropic/claude-3-5-sonnet-20241022"` |
+| `OPENHANDS_MAX_ITER`             | Variable | Set max limit for agent iterations                                                                  | `OPENHANDS_MAX_ITER=10`                            |
+| `OPENHANDS_MACRO`                | Variable | Customize default macro for invoking the resolver                                                   | `OPENHANDS_MACRO=@resolveit`                       |
+| `OPENHANDS_BASE_CONTAINER_IMAGE` | Variable | Custom Sandbox ([learn more](https://docs.all-hands.dev/modules/usage/how-to/custom-sandbox-guide)) | `OPENHANDS_BASE_CONTAINER_IMAGE="custom_image"`    |
+| `TARGET_BRANCH`                  | Variable | Merge to branch other than `main`                                                                   | `TARGET_BRANCH="dev"`                              |

docs/modules/usage/how-to/gui-mode.md

@@ -76,18 +76,18 @@ When using OpenHands in online mode, the GitHub OAuth flow:
 
 Common issues and solutions:
 
-1. **Token Not Recognized**:
+- **Token Not Recognized**:
    - Ensure the token is properly saved in settings.
    - Check that the token hasn't expired.
    - Verify the token has the required scopes.
    - Try regenerating the token.
 
-2. **Organization Access Denied**:
+- **Organization Access Denied**:
    - Check if SSO is required but not enabled.
    - Verify organization membership.
    - Contact organization admin if token policies are blocking access.
 
-3. **Verifying Token Works**:
+- **Verifying Token Works**:
    - The app will show a green checkmark if the token is valid.
    - Try accessing a repository to confirm permissions.
    - Check the browser console for any error messages.

docs/modules/usage/how-to/headless-mode.md

@@ -32,7 +32,7 @@ To run OpenHands in Headless mode with Docker:
 ```bash
 docker run -it \
     --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -e SANDBOX_USER_ID=$(id -u) \
     -e WORKSPACE_MOUNT_PATH=$WORKSPACE_BASE \
     -e LLM_API_KEY=$LLM_API_KEY \
@@ -43,7 +43,7 @@ docker run -it \
     -v ~/.openhands-state:/.openhands-state \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app-$(date +%Y%m%d%H%M%S) \
-    docker.all-hands.dev/all-hands-ai/openhands:0.18 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.24 \
     python -m openhands.core.main -t "write a bash script that prints hi"
 ```
 

docs/modules/usage/how-to/persist-session-data.md

@@ -1,16 +0,0 @@
-# Persisting Session Data
-
-Using the standard Development Workflow, the session data is stored in memory. Currently, if OpenHands' service is restarted,
-previous sessions become invalid (a new secret is generated) and thus not recoverable.
-
-## How to Persist Session Data
-
-### Development Workflow
-In the `config.toml` file, specify the following:
-```
-[core]
-...
-file_store="local"
-file_store_path="/absolute/path/to/openhands/cache/directory"
-jwt_secret="secretpass"
-```

docs/modules/usage/installation.mdx

@@ -1,27 +1,70 @@
-# Installation
+# Running OpenHands
 
 ## System Requirements
 
-- Docker version 26.0.0+ or Docker Desktop 4.31.0+.
-- You must be using Linux or Mac OS.
-  - If you are on Windows, you must use [WSL](https://learn.microsoft.com/en-us/windows/wsl/install).
+- MacOS with [Docker Desktop support](https://docs.docker.com/desktop/setup/install/mac-install/#system-requirements)
+- Linux
+- Windows with [WSL](https://learn.microsoft.com/en-us/windows/wsl/install) and [Docker Desktop support](https://docs.docker.com/desktop/setup/install/windows-install/#system-requirements)
 
-## Start the app
+## Prerequisites
+
+<details>
+  <summary>MacOS</summary>
+  ### Docker Desktop
+
+  1. [Install Docker Desktop on Mac](https://docs.docker.com/desktop/setup/install/mac-install).
+  2. Open Docker Desktop, go to `Settings > Advanced` and ensure `Allow the default Docker socket to be used` is enabled.
+</details>
+
+<details>
+  <summary>Linux</summary>
+
+  :::note
+  Tested with Ubuntu 22.04.
+  :::
+
+  ### Docker Desktop
+
+  1. [Install Docker Desktop on Linux](https://docs.docker.com/desktop/setup/install/linux/).
+
+</details>
+
+<details>
+  <summary>Windows</summary>
+  ### WSL
+
+  1. [Install WSL](https://learn.microsoft.com/en-us/windows/wsl/install).
+  2. Run `wsl --version` in powershell and confirm `Default Version: 2`.
+
+  ### Docker Desktop
+
+  1. [Install Docker Desktop on Windows](https://docs.docker.com/desktop/setup/install/windows-install).
+  2. Open Docker Desktop, go to `Settings` and confirm the following:
+  - General: `Use the WSL 2 based engine` is enabled.
+  - Resources > WSL Integration: `Enable integration with my default WSL distro` is enabled.
+
+  :::note
+  The docker command below to start the app must be run inside the WSL terminal.
+  :::
+
+</details>
+
+## Start the App
 
 The easiest way to run OpenHands is in Docker.
 
 ```bash
-docker pull docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik
 
 docker run -it --rm --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -e LOG_ALL_EVENTS=true \
     -v /var/run/docker.sock:/var/run/docker.sock \
     -v ~/.openhands-state:/.openhands-state \
     -p 3000:3000 \
     --add-host host.docker.internal:host-gateway \
     --name openhands-app \
-    docker.all-hands.dev/all-hands-ai/openhands:0.18
+    docker.all-hands.dev/all-hands-ai/openhands:0.24
 ```
 
 You'll find OpenHands running at http://localhost:3000!

docs/modules/usage/llms/custom-llm-configs.md

@@ -0,0 +1,136 @@
+# Custom LLM Configurations
+
+OpenHands supports defining multiple named LLM configurations in your `config.toml` file. This feature allows you to use different LLM configurations for different purposes, such as using a cheaper model for tasks that don't require high-quality responses, or using different models with different parameters for specific agents.
+
+## How It Works
+
+Named LLM configurations are defined in the `config.toml` file using sections that start with `llm.`. For example:
+
+```toml
+# Default LLM configuration
+[llm]
+model = "gpt-4"
+api_key = "your-api-key"
+temperature = 0.0
+
+# Custom LLM configuration for a cheaper model
+[llm.gpt3]
+model = "gpt-3.5-turbo"
+api_key = "your-api-key"
+temperature = 0.2
+
+# Another custom configuration with different parameters
+[llm.high-creativity]
+model = "gpt-4"
+api_key = "your-api-key"
+temperature = 0.8
+top_p = 0.9
+```
+
+Each named configuration inherits all settings from the default `[llm]` section and can override any of those settings. You can define as many custom configurations as needed.
+
+## Using Custom Configurations
+
+### With Agents
+
+You can specify which LLM configuration an agent should use by setting the `llm_config` parameter in the agent's configuration section:
+
+```toml
+[agent.RepoExplorerAgent]
+# Use the cheaper GPT-3 configuration for this agent
+llm_config = 'gpt3'
+
+[agent.CodeWriterAgent]
+# Use the high creativity configuration for this agent
+llm_config = 'high-creativity'
+```
+
+### Configuration Options
+
+Each named LLM configuration supports all the same options as the default LLM configuration. These include:
+
+- Model selection (`model`)
+- API configuration (`api_key`, `base_url`, etc.)
+- Model parameters (`temperature`, `top_p`, etc.)
+- Retry settings (`num_retries`, `retry_multiplier`, etc.)
+- Token limits (`max_input_tokens`, `max_output_tokens`)
+- And all other LLM configuration options
+
+For a complete list of available options, see the LLM Configuration section in the [Configuration Options](../configuration-options) documentation.
+
+## Use Cases
+
+Custom LLM configurations are particularly useful in several scenarios:
+
+- **Cost Optimization**: Use cheaper models for tasks that don't require high-quality responses, like repository exploration or simple file operations.
+- **Task-Specific Tuning**: Configure different temperature and top_p values for tasks that require different levels of creativity or determinism.
+- **Different Providers**: Use different LLM providers or API endpoints for different tasks.
+- **Testing and Development**: Easily switch between different model configurations during development and testing.
+
+## Example: Cost Optimization
+
+A practical example of using custom LLM configurations to optimize costs:
+
+```toml
+# Default configuration using GPT-4 for high-quality responses
+[llm]
+model = "gpt-4"
+api_key = "your-api-key"
+temperature = 0.0
+
+# Cheaper configuration for repository exploration
+[llm.repo-explorer]
+model = "gpt-3.5-turbo"
+temperature = 0.2
+
+# Configuration for code generation
+[llm.code-gen]
+model = "gpt-4"
+temperature = 0.0
+max_output_tokens = 2000
+
+[agent.RepoExplorerAgent]
+llm_config = 'repo-explorer'
+
+[agent.CodeWriterAgent]
+llm_config = 'code-gen'
+```
+
+In this example:
+- Repository exploration uses a cheaper model since it mainly involves understanding and navigating code
+- Code generation uses GPT-4 with a higher token limit for generating larger code blocks
+- The default configuration remains available for other tasks
+
+# Custom Configurations with Reserved Names
+
+OpenHands can use custom LLM configurations named with reserved names, for specific use cases. If you specify the model and other settings under the reserved names, then OpenHands will load and them for a specific purpose. As of now, one such configuration is implemented: draft editor.
+
+## Draft Editor Configuration
+
+The `draft_editor` configuration is a group of settings you can provide, to specify the model to use for preliminary drafting of code edits, for any tasks that involve editing and refining code. You need to provide it under the section `[llm.draft_editor]`.
+
+For example, you can define in `config.toml` a draft editor like this:
+
+```toml
+[llm.draft_editor]
+model = "gpt-4"
+temperature = 0.2
+top_p = 0.95
+presence_penalty = 0.0
+frequency_penalty = 0.0
+```
+
+This configuration:
+- Uses GPT-4 for high-quality edits and suggestions
+- Sets a low temperature (0.2) to maintain consistency while allowing some flexibility
+- Uses a high top_p value (0.95) to consider a wide range of token options
+- Disables presence and frequency penalties to maintain focus on the specific edits needed
+
+Use this configuration when you want to let an LLM draft edits before making them. In general, it may be useful to:
+- Review and suggest code improvements
+- Refine existing content while maintaining its core meaning
+- Make precise, focused changes to code or text
+
+:::note
+Custom LLM configurations are only available when using OpenHands in development mode, via `main.py` or `cli.py`. When running via `docker run`, please use the standard configuration options.
+:::

docs/modules/usage/llms/llms.md

@@ -5,23 +5,14 @@ OpenHands can connect to any LLM supported by LiteLLM. However, it requires a po
 ## Model Recommendations
 
 Based on our evaluations of language models for coding tasks (using the SWE-bench dataset), we can provide some
-recommendations for model selection. Some analyses can be found in [this blog article comparing LLMs](https://www.all-hands.dev/blog/evaluation-of-llms-as-coding-agents-on-swe-bench-at-30x-speed) and
-[this blog article with some more recent results](https://www.all-hands.dev/blog/openhands-codeact-21-an-open-state-of-the-art-software-development-agent).
-
-When choosing a model, consider both the quality of outputs and the associated costs. Here's a summary of the findings:
-
-- Claude 3.5 Sonnet is the best by a fair amount, achieving a 53% resolve rate on SWE-Bench Verified with the default agent in OpenHands.
-- GPT-4o lags behind, and o1-mini actually performed somewhat worse than GPT-4o. We went in and analyzed the results a little, and briefly it seemed like o1 was sometimes "overthinking" things, performing extra environment configuration tasks when it could just go ahead and finish the task.
-- Finally, the strongest open models were Llama 3.1 405 B and deepseek-v2.5, and they performed reasonably, even besting some of the closed models.
-
-Please refer to the [full article](https://www.all-hands.dev/blog/evaluation-of-llms-as-coding-agents-on-swe-bench-at-30x-speed) for more details.
+recommendations for model selection. Our latest benchmarking results can be found in [this spreadsheet](https://docs.google.com/spreadsheets/d/1wOUdFCMyY6Nt0AIqF705KN4JKOWgeI4wUGUP60krXXs/edit?gid=0).
 
 Based on these findings and community feedback, the following models have been verified to work reasonably well with OpenHands:
 
-- claude-3-5-sonnet (recommended)
-- gpt-4 / gpt-4o
-- llama-3.1-405b
-- deepseek-v2.5
+- anthropic/claude-3-5-sonnet-20241022 (recommended)
+- anthropic/claude-3-5-haiku-20241022
+- deepseek/deepseek-chat
+- gpt-4o
 
 :::warning
 OpenHands will issue many prompts to the LLM you configure. Most of these LLMs cost money, so be sure to set spending

docs/modules/usage/prompting/customization.md

@@ -1,67 +0,0 @@
-# Customizing Agent Behavior
-
-OpenHands can be customized to work more effectively with specific repositories by providing repository-specific context and guidelines. This section explains how to optimize OpenHands for your project.
-
-## Repository Configuration
-
-You can customize OpenHands' behavior for your repository by creating a `.openhands` directory in your repository's root. At minimum, it should contain the file
-`.openhands/microagents/repo.md`, which includes instructions that will
-be given to the agent every time it works with this repository.
-
-We suggest including the following information:
-- **Repository Overview**: A brief description of your project's purpose and architecture.
-- **Directory Structure**: Key directories and their purposes.
-- **Development Guidelines**: Project-specific coding standards and practices.
-- **Testing Requirements**: How to run tests and what types of tests are required.
-- **Setup Instructions**: Steps needed to build and run the project.
-
-### Example Repository Configuration
-Example `.openhands/microagents/repo.md` file:
-```
-Repository: MyProject
-Description: A web application for task management
-
-Directory Structure:
-- src/: Main application code
-- tests/: Test files
-- docs/: Documentation
-
-Setup:
-- Run `npm install` to install dependencies
-- Use `npm run dev` for development
-- Run `npm test` for testing
-
-Guidelines:
-- Follow ESLint configuration
-- Write tests for all new features
-- Use TypeScript for new code
-```
-
-### Customizing Prompts
-
-When working with a repository:
-
-- **Reference Project Standards**: Mention specific coding standards or patterns used in your project.
-- **Include Context**: Reference relevant documentation or existing implementations.
-- **Specify Testing Requirements**: Include project-specific testing requirements in your prompts.
-
-Example customized prompt:
-```
-Add a new task completion feature to src/components/TaskList.tsx following our existing component patterns.
-Include unit tests in tests/components/ and update the documentation in docs/features/.
-The component should use our shared styling from src/styles/components.
-```
-
-### Best Practices for Repository Customization
-
-- **Keep Instructions Updated**: Regularly update your `.openhands` directory as your project evolves.
-- **Be Specific**: Include specific paths, patterns, and requirements unique to your project.
-- **Document Dependencies**: List all tools and dependencies required for development.
-- **Include Examples**: Provide examples of good code patterns from your project.
-- **Specify Conventions**: Document naming conventions, file organization, and code style preferences.
-
-By customizing OpenHands for your repository, you'll get more accurate and consistent results that align with your project's standards and requirements.
-
-## Other Microagents
-You can create other instructions in the `.openhands/microagents/` directory
-that will be sent to the agent if a particular keyword is found, like `test`, `frontend`, or `migration`. See [Micro-Agents](microagents.md) for more information.

docs/modules/usage/prompting/microagents-overview.md

@@ -0,0 +1,36 @@
+# Microagents Overview
+
+Microagents are specialized prompts that enhance OpenHands with domain-specific knowledge, repository-specific context
+and task-specific workflows. They help by providing expert guidance, automating common tasks, and ensuring
+consistent practices across projects.
+
+## Microagent Types
+
+Currently OpenHands supports the following types of microagents:
+
+* [Repository Microagents](./microagents-repo): Repository-specific context and guidelines for OpenHands.
+* [Public Microagents](./microagents-public): General guidelines triggered by keywords for all OpenHands users.
+
+When OpenHands works with a repository, it:
+
+1. Loads repository-specific instructions from `.openhands/microagents/` if present in the repository.
+2. Loads general guidelines triggered by keywords in conversations.
+See current [Public Microagents](https://github.com/All-Hands-AI/OpenHands/tree/main/microagents/knowledge).
+
+## Microagent Format
+
+All microagents use markdown files with YAML frontmatter that have special instructions to help OpenHands accomplish
+tasks:
+```
+---
+name: <Name of the microagent>
+type: <MicroAgent type>
+version: <MicroAgent version>
+agent: <The agent type (Typically CodeActAgent)>
+triggers:
+- <Optional keywords triggering the microagent. If triggers are removed, it will always be included>
+---
+
+<Markdown with any special guidelines, instructions, and prompts that OpenHands should follow.
+Check out the specific documentation for each microagent on best practices for more information.>
+```

docs/modules/usage/prompting/microagents-public.md

@@ -0,0 +1,153 @@
+# Public Microagents
+
+## Overview
+
+Public microagents are specialized guidelines triggered by keywords for all OpenHands users.
+They are defined in markdown files under the
+[`microagents/knowledge/`](https://github.com/All-Hands-AI/OpenHands/tree/main/microagents/knowledge) directory.
+
+Public microagents:
+- Monitor incoming commands for their trigger words.
+- Activate when relevant triggers are detected.
+- Apply their specialized knowledge and capabilities.
+- Follow their specific guidelines and restrictions.
+
+## Current Public Microagents
+
+For more information about specific microagents, refer to their individual documentation files in
+the [`microagents/knowledge/`](https://github.com/All-Hands-AI/OpenHands/tree/main/microagents/knowledge/) directory.
+
+### GitHub Agent
+**File**: `github.md`
+**Triggers**: `github`, `git`
+
+The GitHub agent specializes in GitHub API interactions and repository management. It:
+- Has access to a `GITHUB_TOKEN` for API authentication.
+- Follows strict guidelines for repository interactions.
+- Handles branch management and pull requests.
+- Uses the GitHub API instead of web browser interactions.
+
+Key features:
+- Branch protection (prevents direct pushes to main/master)
+- Automated PR creation
+- Git configuration management
+- API-first approach for GitHub operations
+
+Usage Example:
+
+```bash
+git checkout -b feature-branch
+git commit -m "Add new feature"
+git push origin feature-branch
+```
+
+### NPM Agent
+**File**: `npm.md`
+**Triggers**: `npm`
+
+Specializes in handling npm package management with specific focus on:
+- Non-interactive shell operations.
+- Automated confirmation handling using Unix 'yes' command.
+- Package installation automation.
+
+Usage Example:
+
+```bash
+yes | npm install package-name
+```
+
+## Contributing a Public Microagent
+
+You can create your own public microagents by adding new markdown files to the
+[`microagents/knowledge/`](https://github.com/All-Hands-AI/OpenHands/tree/main/microagents/knowledge/) directory.
+
+### Public Microagents Best Practices
+
+- **Clear Scope**: Keep the microagent focused on a specific domain or task.
+- **Explicit Instructions**: Provide clear, unambiguous guidelines.
+- **Useful Examples**: Include practical examples of common use cases.
+- **Safety First**: Include necessary warnings and constraints.
+- **Integration Awareness**: Consider how the microagent interacts with other components.
+
+### Steps to Contribute a Public Microagent
+
+#### 1. Plan the Public Microagent
+
+Before creating a public microagent, consider:
+- What specific problem or use case will it address?
+- What unique capabilities or knowledge should it have?
+- What trigger words make sense for activating it?
+- What constraints or guidelines should it follow?
+
+#### 2. Create File
+
+Create a new markdown file in [`microagents/knowledge/`](https://github.com/All-Hands-AI/OpenHands/tree/main/microagents/knowledge/)
+with a descriptive name (e.g., `docker.md` for a Docker-focused agent).
+
+Update the file with the required frontmatter [according to the required format](./microagents-overview#microagent-format)
+and the required specialized guidelines while following the [best practices above](#public-microagents-best-practices).
+
+#### 3. Testing the Public Microagent
+
+- Test the agent with various prompts.
+- Verify trigger words activate the agent correctly.
+- Ensure instructions are clear and comprehensive.
+- Check for potential conflicts with existing agents.
+
+#### 4. Submission Process
+
+Submit a pull request with:
+- The new microagent file.
+- Updated documentation if needed.
+- Description of the agent's purpose and capabilities.
+
+### Example Public Microagent Implementation
+
+Here's a template for a new microagent:
+
+```markdown
+---
+name: docker
+agent: CodeActAgent
+triggers:
+- docker
+- container
+---
+
+You are responsible for Docker container management and Dockerfile creation.
+
+Key responsibilities:
+1. Create and modify Dockerfiles
+2. Manage container lifecycle
+3. Handle Docker Compose configurations
+
+Guidelines:
+- Always use official base images when possible
+- Include necessary security considerations
+- Follow Docker best practices for layer optimization
+
+Examples:
+1. Creating a Dockerfile:
+   FROM node:18-alpine
+   WORKDIR /app
+   COPY package*.json ./
+   RUN npm install
+   COPY . .
+   CMD ["npm", "start"]
+
+2. Docker Compose usage:
+   version: '3'
+   services:
+     web:
+       build: .
+       ports:
+         - "3000:3000"
+
+Remember to:
+- Validate Dockerfile syntax
+- Check for security vulnerabilities
+- Optimize for build time and image size
+```
+
+See the [current public micro-agents](https://github.com/All-Hands-AI/OpenHands/tree/main/microagents/knowledge) for
+more examples.

docs/modules/usage/prompting/microagents-repo.md

@@ -0,0 +1,68 @@
+# Repository Microagents
+
+## Overview
+
+OpenHands can be customized to work more effectively with specific repositories by providing repository-specific context
+and guidelines. This section explains how to optimize OpenHands for your project.
+
+## Creating a Repository Micro-Agent
+
+You can customize OpenHands' behavior for your repository by creating a `.openhands/microagents/` directory in your repository's root.
+At minimum it should contain the file
+`.openhands/microagents/repo.md`, which includes instructions that will
+be given to the agent every time it works with this repository.
+
+### Repository Microagents Best Practices
+
+- **Keep Instructions Updated**: Regularly update your `.openhands/microagents/` directory as your project evolves.
+- **Be Specific**: Include specific paths, patterns, and requirements unique to your project.
+- **Document Dependencies**: List all tools and dependencies required for development.
+- **Include Examples**: Provide examples of good code patterns from your project.
+- **Specify Conventions**: Document naming conventions, file organization, and code style preferences.
+
+### Steps to Create a Repository Microagent
+
+#### 1. Plan the Repository Microagent
+When creating a repository-specific micro-agent, we suggest including the following information:
+- **Repository Overview**: A brief description of your project's purpose and architecture.
+- **Directory Structure**: Key directories and their purposes.
+- **Development Guidelines**: Project-specific coding standards and practices.
+- **Testing Requirements**: How to run tests and what types of tests are required.
+- **Setup Instructions**: Steps needed to build and run the project.
+
+#### 2. Create File
+
+Create a file in your repository under `.openhands/microagents/` (Example: `.openhands/microagents/repo.md`)
+
+Update the file with the required frontmatter [according to the required format](./microagents-overview#microagent-format)
+and the required specialized guidelines for your repository.
+
+### Example Repository Microagent
+
+```
+---
+name: repo
+type: repo
+agent: CodeActAgent
+---
+
+Repository: MyProject
+Description: A web application for task management
+
+Directory Structure:
+- src/: Main application code
+- tests/: Test files
+- docs/: Documentation
+
+Setup:
+- Run `npm install` to install dependencies
+- Use `npm run dev` for development
+- Run `npm test` for testing
+
+Guidelines:
+- Follow ESLint configuration
+- Write tests for all new features
+- Use TypeScript for new code
+
+If adding a new component in src/components, always add appropriate unit tests in tests/components/.
+```

docs/modules/usage/prompting/microagents.md

@@ -1,210 +0,0 @@
-# Micro-Agents
-
-OpenHands uses specialized micro-agents to handle specific tasks and contexts efficiently. These micro-agents are small, focused components that provide specialized behavior and knowledge for particular scenarios.
-
-## Overview
-
-Micro-agents are defined in markdown files under the `openhands/agenthub/codeact_agent/micro/` directory. Each micro-agent is configured with:
-
-- A unique name.
-- The agent type (typically CodeActAgent).
-- Trigger keywords that activate the agent.
-- Specific instructions and capabilities.
-
-## Available Micro-Agents
-
-### GitHub Agent
-**File**: `github.md`
-**Triggers**: `github`, `git`
-
-The GitHub agent specializes in GitHub API interactions and repository management. It:
-- Has access to a `GITHUB_TOKEN` for API authentication.
-- Follows strict guidelines for repository interactions.
-- Handles branch management and pull requests.
-- Uses the GitHub API instead of web browser interactions.
-
-Key features:
-- Branch protection (prevents direct pushes to main/master)
-- Automated PR creation
-- Git configuration management
-- API-first approach for GitHub operations
-
-### NPM Agent
-**File**: `npm.md`
-**Triggers**: `npm`
-
-Specializes in handling npm package management with specific focus on:
-- Non-interactive shell operations.
-- Automated confirmation handling using Unix 'yes' command.
-- Package installation automation.
-
-### Custom Micro-Agents
-
-You can create your own micro-agents by adding new markdown files to the micro-agents directory.
-Each file should follow this structure:
-
-```markdown
----
-name: agent_name
-agent: CodeActAgent
-triggers:
-- trigger_word1
-- trigger_word2
----
-
-Instructions and capabilities for the micro-agent...
-```
-
-## Best Practices
-
-When working with micro-agents:
-- **Use Appropriate Triggers**: Ensure your commands include the relevant trigger words to activate the correct micro-agent.
-- **Follow Agent Guidelines**: Each agent has specific instructions and limitations. Respect these for optimal results.
-- **API-First Approach**: When available, use API endpoints rather than web interfaces.
-- **Automation Friendly**: Design commands that work well in non-interactive environments.
-
-## Integration
-
-Micro-agents are automatically integrated into OpenHands' workflow. They:
-- Monitor incoming commands for their trigger words.
-- Activate when relevant triggers are detected.
-- Apply their specialized knowledge and capabilities.
-- Follow their specific guidelines and restrictions.
-
-## Example Usage
-
-```bash
-# GitHub agent example
-git checkout -b feature-branch
-git commit -m "Add new feature"
-git push origin feature-branch
-
-# NPM agent example
-yes | npm install package-name
-```
-
-For more information about specific agents, refer to their individual documentation files in the micro-agents directory.
-
-## Contributing a Micro-Agent
-
-To contribute a new micro-agent to OpenHands, follow these guidelines:
-
-### 1. Planning Your Micro-Agent
-
-Before creating a micro-agent, consider:
-- What specific problem or use case will it address?
-- What unique capabilities or knowledge should it have?
-- What trigger words make sense for activating it?
-- What constraints or guidelines should it follow?
-
-### 2. File Structure
-
-Create a new markdown file in `openhands/agenthub/codeact_agent/micro/` with a descriptive name (e.g., `docker.md` for a Docker-focused agent).
-
-### 3. Required Components
-
-Your micro-agent file must include:
-
-- **Front Matter**: YAML metadata at the start of the file:
-```markdown
----
-name: your_agent_name
-agent: CodeActAgent
-triggers:
-- trigger_word1
-- trigger_word2
----
-```
-
-- **Instructions**: Clear, specific guidelines for the agent's behavior:
-```markdown
-You are responsible for [specific task/domain].
-
-Key responsibilities:
-1. [Responsibility 1]
-2. [Responsibility 2]
-
-Guidelines:
-- [Guideline 1]
-- [Guideline 2]
-
-Examples of usage:
-[Example 1]
-[Example 2]
-```
-
-### 4. Best Practices for Micro-Agent Development
-
-- **Clear Scope**: Keep the agent focused on a specific domain or task.
-- **Explicit Instructions**: Provide clear, unambiguous guidelines.
-- **Useful Examples**: Include practical examples of common use cases.
-- **Safety First**: Include necessary warnings and constraints.
-- **Integration Awareness**: Consider how the agent interacts with other components.
-
-### 5. Testing Your Micro-Agent
-
-Before submitting:
-- Test the agent with various prompts.
-- Verify trigger words activate the agent correctly.
-- Ensure instructions are clear and comprehensive.
-- Check for potential conflicts with existing agents.
-
-### 6. Example Implementation
-
-Here's a template for a new micro-agent:
-
-```markdown
----
-name: docker
-agent: CodeActAgent
-triggers:
-- docker
-- container
----
-
-You are responsible for Docker container management and Dockerfile creation.
-
-Key responsibilities:
-1. Create and modify Dockerfiles
-2. Manage container lifecycle
-3. Handle Docker Compose configurations
-
-Guidelines:
-- Always use official base images when possible
-- Include necessary security considerations
-- Follow Docker best practices for layer optimization
-
-Examples:
-1. Creating a Dockerfile:
-   FROM node:18-alpine
-   WORKDIR /app
-   COPY package*.json ./
-   RUN npm install
-   COPY . .
-   CMD ["npm", "start"]
-
-2. Docker Compose usage:
-   version: '3'
-   services:
-     web:
-       build: .
-       ports:
-         - "3000:3000"
-
-Remember to:
-- Validate Dockerfile syntax
-- Check for security vulnerabilities
-- Optimize for build time and image size
-```
-
-### 7. Submission Process
-
-1. Create your micro-agent file in the correct directory.
-2. Test thoroughly.
-3. Submit a pull request with:
-   - The new micro-agent file.
-   - Updated documentation if needed.
-   - Description of the agent's purpose and capabilities.
-
-Remember that micro-agents are a powerful way to extend OpenHands' capabilities in specific domains. Well-designed
-agents can significantly improve the system's ability to handle specialized tasks.

docs/modules/usage/runtimes.md

@@ -16,7 +16,7 @@ some flags being passed to `docker run` that make this possible:
 
 ```
 docker run # ...
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.18-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.24-nikolaik \
     -v /var/run/docker.sock:/var/run/docker.sock \
     # ...
 ```

docs/package.json

@@ -15,23 +15,23 @@
     "typecheck": "tsc"
   },
   "dependencies": {
-    "@docusaurus/core": "^3.6.3",
-    "@docusaurus/plugin-content-pages": "^3.6.3",
-    "@docusaurus/preset-classic": "^3.6.3",
-    "@docusaurus/theme-mermaid": "^3.6.3",
+    "@docusaurus/core": "^3.7.0",
+    "@docusaurus/plugin-content-pages": "^3.7.0",
+    "@docusaurus/preset-classic": "^3.7.0",
+    "@docusaurus/theme-mermaid": "^3.7.0",
     "@mdx-js/react": "^3.1.0",
     "clsx": "^2.0.0",
     "prism-react-renderer": "^2.4.1",
-    "react": "^18.3.1",
-    "react-dom": "^18.3.1",
+    "react": "^19.0.0",
+    "react-dom": "^19.0.0",
     "react-icons": "^5.4.0",
     "react-use": "^17.6.0"
   },
   "devDependencies": {
     "@docusaurus/module-type-aliases": "^3.5.1",
-    "@docusaurus/tsconfig": "^3.6.3",
+    "@docusaurus/tsconfig": "^3.7.0",
     "@docusaurus/types": "^3.5.1",
-    "typescript": "~5.7.2"
+    "typescript": "~5.7.3"
   },
   "browserslist": {
     "production": [

docs/sidebars.ts

@@ -5,7 +5,7 @@ const sidebars: SidebarsConfig = {
   docsSidebar: [
     {
       type: 'doc',
-      label: 'Installation',
+      label: 'Running OpenHands',
       id: 'usage/installation',
     },
     {
@@ -23,15 +23,26 @@ const sidebars: SidebarsConfig = {
           id: 'usage/prompting/prompting-best-practices',
         },
         {
-          type: 'doc',
-          label: 'Customization',
-          id: 'usage/prompting/customization',
-        },
-        {
-          type: 'doc',
+          type: 'category',
           label: 'Microagents',
-          id: 'usage/prompting/microagents',
-        },
+          items: [
+            {
+              type: 'doc',
+              label: 'Overview',
+              id: 'usage/prompting/microagents-overview',
+            },
+            {
+              type: 'doc',
+              label: 'Repository',
+              id: 'usage/prompting/microagents-repo',
+            },
+            {
+              type: 'doc',
+              label: 'Public',
+              id: 'usage/prompting/microagents-public',
+            },
+          ],
+        }
       ],
     },
     {
@@ -126,11 +137,6 @@ const sidebars: SidebarsConfig = {
           label: 'Custom Sandbox',
           id: 'usage/how-to/custom-sandbox-guide',
         },
-        {
-          type: 'doc',
-          label: 'Persist Session Data',
-          id: 'usage/how-to/persist-session-data',
-        },
       ],
     },
     {

docs/src/components/CustomFooter.tsx

@@ -8,7 +8,7 @@ function CustomFooter() {
     <footer className="custom-footer">
       <div className="footer-content">
         <div className="footer-icons">
-          <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2wkh4pklz-w~h_DVDtEe9H5kyQlcNxVw" target="_blank" rel="noopener noreferrer">
+          <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg" target="_blank" rel="noopener noreferrer">
             <FaSlack />
           </a>
           <a href="https://discord.gg/ESHStjSjD4" target="_blank" rel="noopener noreferrer">

docs/src/components/HomepageHeader/HomepageHeader.tsx

@@ -23,7 +23,7 @@ export function HomepageHeader() {
           <a href="https://codecov.io/github/All-Hands-AI/OpenHands?branch=main"><img alt="CodeCov" src="https://img.shields.io/codecov/c/github/All-Hands-AI/OpenHands?style=for-the-badge&color=blue" /></a>
           <a href="https://github.com/All-Hands-AI/OpenHands/blob/main/LICENSE"><img src="https://img.shields.io/github/license/All-Hands-AI/OpenHands?style=for-the-badge&color=blue" alt="MIT License" /></a>
           <br/>
-          <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2wkh4pklz-w~h_DVDtEe9H5kyQlcNxVw"><img src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge" alt="Join our Slack community" /></a>
+          <a href="https://join.slack.com/t/openhands-ai/shared_invite/zt-2ypg5jweb-d~6hObZDbXi_HEL8PDrbHg"><img src="https://img.shields.io/badge/Slack-Join%20Us-red?logo=slack&logoColor=white&style=for-the-badge" alt="Join our Slack community" /></a>
           <a href="https://discord.gg/ESHStjSjD4"><img src="https://img.shields.io/badge/Discord-Join%20Us-purple?logo=discord&logoColor=white&style=for-the-badge" alt="Join our Discord community" /></a>
           <a href="https://github.com/All-Hands-AI/OpenHands/blob/main/CREDITS.md"><img src="https://img.shields.io/badge/Project-Credits-blue?style=for-the-badge&color=FFE165&logo=github&logoColor=white" alt="Credits" /></a>
           <br/>

evaluation/benchmarks/EDA/run_infer.py

@@ -69,6 +69,7 @@ def get_config(
             base_container_image='python:3.12-bookworm',
             enable_auto_lint=False,
             use_host_network=False,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -76,7 +77,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/agent_bench/run_infer.py

@@ -53,6 +53,7 @@ def get_config(
             remote_runtime_api_url=os.environ.get('SANDBOX_REMOTE_RUNTIME_API_URL'),
             keep_runtime_alive=False,
             remote_runtime_init_timeout=3600,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -60,7 +61,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/aider_bench/run_infer.py

@@ -61,6 +61,7 @@ def get_config(
             remote_runtime_api_url=os.environ.get('SANDBOX_REMOTE_RUNTIME_API_URL'),
             keep_runtime_alive=False,
             remote_runtime_init_timeout=1800,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -68,7 +69,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
 
     # copy 'draft_editor' config if exists
     config_copy = copy.deepcopy(config)

evaluation/benchmarks/biocoder/run_infer.py

@@ -67,6 +67,7 @@ def get_config(
             base_container_image=BIOCODER_BENCH_CONTAINER_IMAGE,
             enable_auto_lint=True,
             use_host_network=False,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -74,7 +75,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/bird/run_infer.py

@@ -80,6 +80,7 @@ def get_config(
             base_container_image='python:3.12-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -87,7 +88,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/browsing_delegation/run_infer.py

@@ -45,13 +45,14 @@ def get_config(
             base_container_image='python:3.12-bookworm',
             enable_auto_lint=False,
             use_host_network=False,
+            remote_runtime_enable_retries=True,
         ),
         workspace_base=None,
         workspace_mount_path=None,
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/commit0_bench/run_infer.py

@@ -135,6 +135,7 @@ def get_config(
             remote_runtime_api_url=os.environ.get('SANDBOX_REMOTE_RUNTIME_API_URL'),
             keep_runtime_alive=False,
             remote_runtime_init_timeout=3600,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -171,7 +172,7 @@ def initialize_runtime(
     action = CmdRunAction(
         command=f'git clone -b commit0_combined https://github.com/{instance["repo"]}.git'
     )
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -181,7 +182,7 @@ def initialize_runtime(
     )
 
     action = CmdRunAction(command=f'cd /workspace/{workspace_dir_name}')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -191,7 +192,7 @@ def initialize_runtime(
     )
 
     action = CmdRunAction(command='git checkout -b openhands')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -201,7 +202,7 @@ def initialize_runtime(
 
     # Install commit0
     action = CmdRunAction(command='/root/.cargo/bin/uv pip install commit0')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     # logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -231,7 +232,7 @@ def complete_runtime(
     workspace_dir_name = _get_commit0_workspace_dir_name(instance)
 
     action = CmdRunAction(command='git add .')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -241,7 +242,7 @@ def complete_runtime(
     )
 
     action = CmdRunAction(command='git commit -m "openhands edits"')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -258,7 +259,7 @@ def complete_runtime(
         action = CmdRunAction(
             command=f"git diff {instance['base_commit']} HEAD -- . ':(exclude)spec.pdf.bz2'"
         )
-        action.timeout = 600 + 100 * n_retries
+        action.set_hard_timeout(600 + 100 * n_retries)
         logger.info(action, extra={'msg_type': 'ACTION'})
         obs = runtime.run_action(action)
         # logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -282,7 +283,7 @@ def complete_runtime(
     action = CmdRunAction(
         command=f"{instance['test']['test_cmd']} --json-report --json-report-file=report.json --continue-on-collection-errors {test_dir} > test_output.txt 2>&1"
     )
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -292,7 +293,7 @@ def complete_runtime(
     )
     # Read test output
     action = CmdRunAction(command='cat test_output.txt')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     # logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -305,7 +306,7 @@ def complete_runtime(
 
     # Save pytest exit code
     action = CmdRunAction(command='echo $?')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     # logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -318,7 +319,7 @@ def complete_runtime(
 
     # Read the test report
     action = CmdRunAction(command='cat report.json')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     # logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -330,7 +331,7 @@ def complete_runtime(
     repo_name = instance['repo'].split('/')[1]
     repo_name = repo_name.replace('.', '-')
     action = CmdRunAction(command=f'commit0 get-tests {repo_name}')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     # logger.info(obs, extra={'msg_type': 'OBSERVATION'})

evaluation/benchmarks/discoverybench/run_infer.py

@@ -71,6 +71,7 @@ def get_config(
             base_container_image='python:3.12-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -78,7 +79,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     agent_config = AgentConfig(
         function_calling=False,
         codeact_enable_jupyter=True,

evaluation/benchmarks/gaia/run_infer.py

@@ -25,6 +25,7 @@ from openhands.core.config import (
     get_llm_config_arg,
     get_parser,
 )
+from openhands.core.config.utils import get_agent_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 AgentFinishAction, CmdRunAction, MessageAction
@@ -56,14 +57,19 @@ def get_config(
             base_container_image='python:3.12-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
         workspace_mount_path=None,
     )
     config.set_llm_config(metadata.llm_config)
-    agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    if metadata.agent_config:
+        config.set_agent_config(metadata.agent_config, metadata.agent_class)
+    else:
+        logger.info('Agent config not provided, using default settings')
+        agent_config = config.get_agent_config(metadata.agent_class)
+        agent_config.enable_prompt_extensions = False
     return config
 
 
@@ -237,6 +243,10 @@ if __name__ == '__main__':
     )
     args, _ = parser.parse_known_args()
 
+    agent_config = None
+    if args.agent_config:
+        agent_config = get_agent_config_arg(args.agent_config)
+
     llm_config = None
     if args.llm_config:
         llm_config = get_llm_config_arg(args.llm_config)
@@ -255,6 +265,7 @@ if __name__ == '__main__':
         eval_output_dir=args.eval_output_dir,
         data_split=args.data_split,
         details={'gaia-level': args.level},
+        agent_config=agent_config,
     )
 
     dataset = load_dataset('gaia-benchmark/GAIA', args.level)

evaluation/benchmarks/gaia/scripts/run_infer.sh

@@ -9,6 +9,7 @@ AGENT=$3
 EVAL_LIMIT=$4
 LEVELS=$5
 NUM_WORKERS=$6
+AGENT_CONFIG=$7
 
 if [ -z "$NUM_WORKERS" ]; then
   NUM_WORKERS=1
@@ -49,5 +50,9 @@ if [ -n "$EVAL_LIMIT" ]; then
   COMMAND="$COMMAND --eval-n-limit $EVAL_LIMIT"
 fi
 
+if [ -n "$AGENT_CONFIG" ]; then
+  echo "AGENT_CONFIG: $AGENT_CONFIG"
+  COMMAND="$COMMAND --agent-config $AGENT_CONFIG"
+
 # Run the command
 eval $COMMAND

evaluation/benchmarks/gorilla/run_infer.py

@@ -49,6 +49,7 @@ def get_config(
             base_container_image='python:3.12-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -56,7 +57,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/gpqa/run_infer.py

@@ -70,6 +70,7 @@ def get_config(
             base_container_image='python:3.12-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -77,7 +78,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/humanevalfix/run_infer.py

@@ -91,6 +91,7 @@ def get_config(
             base_container_image='python:3.12-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -98,7 +99,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/logic_reasoning/run_infer.py

@@ -55,6 +55,7 @@ def get_config(
             enable_auto_lint=True,
             use_host_network=False,
             runtime_extra_deps='$OH_INTERPRETER_PATH -m pip install scitools-pyke',
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -62,7 +63,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/miniwob/README.md

@@ -8,6 +8,9 @@ Please follow instruction [here](../../README.md#setup) to setup your local deve
 
 ## Test if your environment works
 
+Follow the instructions here https://miniwob.farama.org/content/getting_started/ & https://miniwob.farama.org/content/viewing/
+to set up MiniWoB server in your local environment at http://localhost:8080/miniwob/
+
 Access with browser the above MiniWoB URLs and see if they load correctly.
 
 ## Run Evaluation

evaluation/benchmarks/miniwob/run_infer.py

@@ -70,6 +70,7 @@ def get_config(
             remote_runtime_init_timeout=1800,
             keep_runtime_alive=False,
             timeout=120,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,

evaluation/benchmarks/mint/run_infer.py

@@ -113,6 +113,7 @@ def get_config(
             enable_auto_lint=True,
             use_host_network=False,
             runtime_extra_deps=f'$OH_INTERPRETER_PATH -m pip install {" ".join(MINT_DEPENDENCIES)}',
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -120,7 +121,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/ml_bench/run_infer.py

@@ -93,7 +93,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/scienceagentbench/run_infer.py

@@ -73,6 +73,7 @@ def get_config(
             api_key=os.environ.get('ALLHANDS_API_KEY', None),
             remote_runtime_api_url=os.environ.get('SANDBOX_REMOTE_RUNTIME_API_URL'),
             keep_runtime_alive=False,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,

evaluation/benchmarks/swe_bench/README.md

@@ -204,7 +204,7 @@ Then, in a separate Python environment with `streamlit` library, you can run the
 ```bash
 # Make sure you are inside the cloned `evaluation` repo
 conda activate streamlit # if you follow the optional conda env setup above
-streamlit app.py --server.port 8501 --server.address 0.0.0.0
+streamlit run app.py --server.port 8501 --server.address 0.0.0.0
 ```
 
 Then you can access the SWE-Bench trajectory visualizer at `localhost:8501`.

evaluation/benchmarks/swe_bench/eval_infer.py

@@ -1,3 +1,4 @@
+import json
 import os
 import tempfile
 import time
@@ -11,7 +12,11 @@ from swebench.harness.run_evaluation import (
 )
 from swebench.harness.test_spec import SWEbenchInstance, TestSpec, make_test_spec
 from swebench.harness.utils import load_swebench_dataset
+from tqdm import tqdm
 
+from evaluation.benchmarks.swe_bench.resource.mapping import (
+    get_instance_resource_factor,
+)
 from evaluation.benchmarks.swe_bench.run_infer import get_instance_docker_image
 from evaluation.utils.shared import (
     EvalMetadata,
@@ -66,7 +71,7 @@ def process_git_patch(patch):
     return patch
 
 
-def get_config(instance: pd.Series) -> AppConfig:
+def get_config(metadata: EvalMetadata, instance: pd.Series) -> AppConfig:
     # We use a different instance image for the each instance of swe-bench eval
     base_container_image = get_instance_docker_image(instance['instance_id'])
     logger.info(
@@ -81,10 +86,14 @@ def get_config(instance: pd.Series) -> AppConfig:
             base_container_image=base_container_image,
             use_host_network=False,
             # large enough timeout, since some testcases take very long to run
-            timeout=1800,
+            timeout=600,
             api_key=os.environ.get('ALLHANDS_API_KEY', None),
             remote_runtime_api_url=os.environ.get('SANDBOX_REMOTE_RUNTIME_API_URL'),
             remote_runtime_init_timeout=3600,
+            remote_runtime_resource_factor=get_instance_resource_factor(
+                dataset_name=metadata.dataset,
+                instance_id=instance['instance_id'],
+            ),
         ),
         # do not mount workspace
         workspace_base=None,
@@ -123,7 +132,7 @@ def process_instance(
     else:
         logger.info(f'Starting evaluation for instance {instance.instance_id}.')
 
-    config = get_config(instance)
+    config = get_config(metadata, instance)
     instance_id = instance.instance_id
     model_patch = instance['model_patch']
     test_spec: TestSpec = instance['test_spec']
@@ -151,52 +160,52 @@ def process_instance(
     if runtime_failure_count > 0:
         config.sandbox.remote_runtime_resource_factor = min(
             config.sandbox.remote_runtime_resource_factor * (2**runtime_failure_count),
-            4,  # hardcode maximum resource factor to 4
+            8,
         )
         logger.warning(
-            f'This is the second attempt for instance {instance.instance_id}, setting resource factor to {config.sandbox.remote_runtime_resource_factor}'
+            f'This is the {runtime_failure_count + 1}th attempt for instance {instance.instance_id}, setting resource factor to {config.sandbox.remote_runtime_resource_factor}'
         )
 
-    runtime = create_runtime(config)
-    call_async_from_sync(runtime.connect)
-    # Get patch and save it to /tmp/patch.diff
-    with tempfile.TemporaryDirectory() as temp_dir:
-        # Patch file
-        patch_file_path = os.path.join(temp_dir, 'patch.diff')
-        with open(patch_file_path, 'w') as f:
-            f.write(model_patch)
-        runtime.copy_to(patch_file_path, '/tmp')
-        # Eval script
-        eval_script_path = os.path.join(temp_dir, 'eval.sh')
-        with open(eval_script_path, 'w') as f:
-            f.write(test_spec.eval_script)
-        runtime.copy_to(eval_script_path, '/tmp')
-
-    # Set +x
-    action = CmdRunAction(command='chmod +x /tmp/eval.sh')
-    action.timeout = 600
-    logger.info(action, extra={'msg_type': 'ACTION'})
-    obs = runtime.run_action(action)
-    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
-    assert obs.exit_code == 0
-
-    # Apply patch
-    exec_command = (
-        'cd /testbed && '
-        "(git apply -v /tmp/patch.diff && echo 'APPLY_PATCH_PASS' || "
-        "(echo 'Failed to apply patch with git apply, trying with patch command...' && "
-        "(patch --batch --fuzz=5 -p1 -i /tmp/patch.diff && echo 'APPLY_PATCH_PASS' || "
-        "echo 'APPLY_PATCH_FAIL')))"
-    )
-    action = CmdRunAction(command=exec_command)
-    action.timeout = 600
-    obs = runtime.run_action(action)
-    assert isinstance(obs, CmdOutputObservation)
-    apply_patch_output = obs.content
-    assert isinstance(apply_patch_output, str)
-    instance['test_result']['apply_patch_output'] = apply_patch_output
-
     try:
+        runtime = create_runtime(config)
+        call_async_from_sync(runtime.connect)
+        # Get patch and save it to /tmp/patch.diff
+        with tempfile.TemporaryDirectory() as temp_dir:
+            # Patch file
+            patch_file_path = os.path.join(temp_dir, 'patch.diff')
+            with open(patch_file_path, 'w') as f:
+                f.write(model_patch)
+            runtime.copy_to(patch_file_path, '/tmp')
+            # Eval script
+            eval_script_path = os.path.join(temp_dir, 'eval.sh')
+            with open(eval_script_path, 'w') as f:
+                f.write(test_spec.eval_script)
+            runtime.copy_to(eval_script_path, '/tmp')
+
+        # Set +x
+        action = CmdRunAction(command='chmod +x /tmp/eval.sh')
+        action.set_hard_timeout(600)
+        logger.info(action, extra={'msg_type': 'ACTION'})
+        obs = runtime.run_action(action)
+        logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+        assert obs.exit_code == 0
+
+        # Apply patch
+        exec_command = (
+            'cd /testbed && '
+            "(git apply -v /tmp/patch.diff && echo 'APPLY_PATCH_PASS' || "
+            "(echo 'Failed to apply patch with git apply, trying with patch command...' && "
+            "(patch --batch --fuzz=5 -p1 -i /tmp/patch.diff && echo 'APPLY_PATCH_PASS' || "
+            "echo 'APPLY_PATCH_FAIL')))"
+        )
+        action = CmdRunAction(command=exec_command)
+        action.set_hard_timeout(600)
+        obs = runtime.run_action(action)
+        assert isinstance(obs, CmdOutputObservation)
+        apply_patch_output = obs.content
+        assert isinstance(apply_patch_output, str)
+        instance['test_result']['apply_patch_output'] = apply_patch_output
+
         if 'APPLY_PATCH_FAIL' in apply_patch_output:
             logger.info(f'[{instance_id}] {APPLY_PATCH_FAIL}:\n{apply_patch_output}')
             instance['test_result']['report']['failed_apply_patch'] = True
@@ -212,7 +221,7 @@ def process_instance(
             # Run eval script in background and save output to log file
             log_file = '/tmp/eval_output.log'
             action = CmdRunAction(command=f'/tmp/eval.sh > {log_file} 2>&1 & echo $!')
-            action.timeout = 60  # Short timeout just to get the process ID
+            action.set_hard_timeout(300)  # Short timeout just to get the process ID
             obs = runtime.run_action(action)
 
             if isinstance(obs, CmdOutputObservation) and obs.exit_code == 0:
@@ -235,7 +244,7 @@ def process_instance(
                     check_action = CmdRunAction(
                         command=f'ps -p {pid} > /dev/null; echo $?'
                     )
-                    check_action.timeout = 60
+                    check_action.set_hard_timeout(300)
                     check_obs = runtime.run_action(check_action)
                     if (
                         isinstance(check_obs, CmdOutputObservation)
@@ -252,7 +261,7 @@ def process_instance(
 
                 # Read the log file
                 cat_action = CmdRunAction(command=f'cat {log_file}')
-                cat_action.timeout = 300
+                cat_action.set_hard_timeout(300)
                 cat_obs = runtime.run_action(cat_action)
 
                 # Grade answer
@@ -352,7 +361,14 @@ if __name__ == '__main__':
 
     # Load predictions
     assert args.input_file.endswith('.jsonl'), 'Input file must be a jsonl file.'
-    predictions = pd.read_json(args.input_file, lines=True)
+    required_fields = ['instance_id', 'model_patch', 'test_result']
+    with open(args.input_file) as f:
+        predictions = pd.DataFrame.from_records(
+            [
+                {k: v for k, v in json.loads(line).items() if k in required_fields}
+                for line in tqdm(f, desc='Loading predictions')
+            ]
+        )
     assert (
         'instance_id' in predictions.columns
     ), 'Input file must contain instance_id column.'
@@ -396,6 +412,17 @@ if __name__ == '__main__':
         with open(metadata_filepath, 'r') as metadata_file:
             data = metadata_file.read()
             metadata = EvalMetadata.model_validate_json(data)
+    else:
+        # Initialize with a dummy metadata when file doesn't exist
+        metadata = EvalMetadata(
+            agent_class="dummy_agent",  # Placeholder agent class
+            llm_config=LLMConfig(model="dummy_model"),  # Minimal LLM config
+            max_iterations=1,  # Minimal iterations
+            eval_output_dir=os.path.dirname(args.input_file),  # Use input file dir as output dir
+            start_time=time.strftime('%Y-%m-%d %H:%M:%S'),  # Current time
+            git_commit=subprocess.check_output(['git', 'rev-parse', 'HEAD']).decode('utf-8').strip(),  # Current commit
+            dataset=args.dataset  # Dataset name from args
+        )
 
     # The evaluation harness constrains the signature of `process_instance_func` but we need to
     # pass extra information. Build a new function object to avoid issues with multiprocessing.

evaluation/benchmarks/swe_bench/resource/mapping.py

@@ -0,0 +1,39 @@
+"""Mapping instance_id to resource_factor.
+
+Different instances may have different resource requirements.
+e.g., some instances may require more memory/CPU to run inference.
+This file tracks the resource requirements of different instances.
+"""
+
+import json
+import os
+
+from openhands.core.logger import openhands_logger as logger
+
+CUR_DIR = os.path.dirname(os.path.abspath(__file__))
+DEFAULT_RUNTIME_RESOURCE_FACTOR = int(
+    os.environ.get('DEFAULT_RUNTIME_RESOURCE_FACTOR', 1)
+)
+
+# dataset to resource mapping
+_global_resource_mapping: dict[str, dict[str, float]] = {}
+
+
+def get_resource_mapping(dataset_name: str) -> dict[str, float]:
+    if dataset_name not in _global_resource_mapping:
+        file_path = os.path.join(CUR_DIR, f'{dataset_name}.json')
+        if not os.path.exists(file_path):
+            logger.warning(f'Resource mapping for {dataset_name} not found.')
+            return None
+
+        with open(file_path, 'r') as f:
+            _global_resource_mapping[dataset_name] = json.load(f)
+        logger.info(f'Loaded resource mapping for {dataset_name}')
+    return _global_resource_mapping[dataset_name]
+
+
+def get_instance_resource_factor(dataset_name: str, instance_id: str) -> int:
+    resource_mapping = get_resource_mapping(dataset_name)
+    if resource_mapping is None:
+        return DEFAULT_RUNTIME_RESOURCE_FACTOR
+    return int(resource_mapping.get(instance_id, DEFAULT_RUNTIME_RESOURCE_FACTOR))

evaluation/benchmarks/swe_bench/resource/princeton-nlp__SWE-bench_Verified-test.json

@@ -0,0 +1 @@
+{"pydata__xarray-6721": 8, "pytest-dev__pytest-7236": 8, "matplotlib__matplotlib-24627": 4, "django__django-15561": 4, "django__django-15098": 4, "django__django-14771": 4, "sympy__sympy-21612": 4, "sympy__sympy-15345": 4, "psf__requests-5414": 4, "astropy__astropy-14508": 2, "django__django-11451": 2, "django__django-11477": 2, "django__django-10880": 2, "django__django-11163": 2, "django__django-11815": 2, "astropy__astropy-14369": 2, "django__django-10097": 2, "django__django-10554": 2, "django__django-12304": 2, "django__django-12325": 2, "django__django-11551": 2, "django__django-11734": 2, "django__django-13109": 2, "django__django-13089": 2, "django__django-13343": 2, "django__django-13363": 2, "django__django-13809": 2, "django__django-13810": 2, "django__django-13786": 2, "django__django-13807": 2, "django__django-14493": 2, "django__django-11820": 2, "django__django-11951": 2, "django__django-11964": 2, "astropy__astropy-14309": 2, "astropy__astropy-14365": 2, "astropy__astropy-12907": 2, "astropy__astropy-14182": 2, "django__django-15161": 2, "django__django-15128": 2, "django__django-14999": 2, "django__django-14915": 2, "django__django-14752": 2, "django__django-14765": 2, "django__django-14089": 2, "django__django-15252": 2, "django__django-15380": 2, "django__django-15382": 2, "django__django-15499": 2, "django__django-15467": 2, "django__django-15280": 2, "django__django-15315": 2, "django__django-15277": 2, "django__django-15268": 2, "django__django-15629": 2, "django__django-15695": 2, "django__django-15732": 2, "django__django-15863": 2, "django__django-16082": 2, "django__django-16145": 2, "django__django-16256": 2, "django__django-16429": 2, "django__django-16454": 2, "django__django-16493": 2, "matplotlib__matplotlib-13989": 2, "matplotlib__matplotlib-20488": 2, "django__django-15503": 2, "django__django-15525": 2, "django__django-15375": 2, "django__django-15278": 2, "matplotlib__matplotlib-21568": 2, "matplotlib__matplotlib-20859": 2, "matplotlib__matplotlib-20826": 2, "matplotlib__matplotlib-20676": 2, "matplotlib__matplotlib-23412": 2, "matplotlib__matplotlib-22719": 2, "matplotlib__matplotlib-23299": 2, "matplotlib__matplotlib-22865": 2, "matplotlib__matplotlib-24149": 2, "matplotlib__matplotlib-24177": 2, "matplotlib__matplotlib-24570": 2, "matplotlib__matplotlib-24637": 2, "matplotlib__matplotlib-24970": 2, "matplotlib__matplotlib-23476": 2, "matplotlib__matplotlib-24026": 2, "matplotlib__matplotlib-23314": 2, "matplotlib__matplotlib-25332": 2, "matplotlib__matplotlib-25311": 2, "matplotlib__matplotlib-25122": 2, "matplotlib__matplotlib-25479": 2, "matplotlib__matplotlib-26342": 2, "psf__requests-2317": 2, "matplotlib__matplotlib-25960": 2, "matplotlib__matplotlib-25775": 2, "pydata__xarray-4356": 2, "pydata__xarray-4075": 2, "pydata__xarray-6461": 2, "pydata__xarray-4687": 2, "pydata__xarray-6599": 2, "pylint-dev__pylint-4661": 2, "django__django-15554": 2, "django__django-15563": 2, "pytest-dev__pytest-5262": 2, "pytest-dev__pytest-10081": 2, "scikit-learn__scikit-learn-12973": 2, "scikit-learn__scikit-learn-13124": 2, "scikit-learn__scikit-learn-13779": 2, "scikit-learn__scikit-learn-14141": 2, "scikit-learn__scikit-learn-13439": 2, "scikit-learn__scikit-learn-13496": 2, "scikit-learn__scikit-learn-15100": 2, "scikit-learn__scikit-learn-25102": 2, "scikit-learn__scikit-learn-25232": 2, "scikit-learn__scikit-learn-25747": 2, "scikit-learn__scikit-learn-26323": 2, "scikit-learn__scikit-learn-9288": 2, "scikit-learn__scikit-learn-14496": 2, "scikit-learn__scikit-learn-14629": 2, "sphinx-doc__sphinx-8265": 2, "sphinx-doc__sphinx-8548": 2, "sphinx-doc__sphinx-8593": 2, "sphinx-doc__sphinx-8595": 2, "sphinx-doc__sphinx-8621": 2, "sphinx-doc__sphinx-8638": 2, "sphinx-doc__sphinx-9229": 2, "sphinx-doc__sphinx-9281": 2, "sphinx-doc__sphinx-9461": 2, "sphinx-doc__sphinx-9591": 2, "sphinx-doc__sphinx-9658": 2, "sphinx-doc__sphinx-9673": 2, "sympy__sympy-12096": 2, "sympy__sympy-12481": 2, "sphinx-doc__sphinx-10323": 2, "sphinx-doc__sphinx-7590": 2, "sympy__sympy-13877": 2, "sympy__sympy-12489": 2, "sympy__sympy-15809": 2, "sympy__sympy-14711": 2, "sympy__sympy-16597": 2, "sympy__sympy-16766": 2, "sympy__sympy-16792": 2, "sympy__sympy-15875": 2, "sympy__sympy-17655": 2, "sympy__sympy-18189": 2, "sympy__sympy-18763": 2, "sympy__sympy-19040": 2, "sympy__sympy-19495": 2, "sympy__sympy-19637": 2, "sympy__sympy-19783": 2, "sympy__sympy-17630": 2, "sympy__sympy-20428": 2, "sympy__sympy-20590": 2, "sympy__sympy-20801": 2, "sympy__sympy-21379": 2, "sympy__sympy-21847": 2, "sympy__sympy-22456": 2, "sympy__sympy-22714": 2, "sympy__sympy-22914": 2, "sympy__sympy-23262": 2, "sympy__sympy-23413": 2, "sympy__sympy-23534": 2, "sympy__sympy-24066": 2, "sympy__sympy-24213": 2, "sympy__sympy-24443": 2, "sympy__sympy-24562": 2, "sympy__sympy-24661": 2}

evaluation/benchmarks/swe_bench/run_infer.py

@@ -9,12 +9,16 @@ import toml
 from datasets import load_dataset
 
 import openhands.agenthub
+from evaluation.benchmarks.swe_bench.resource.mapping import (
+    get_instance_resource_factor,
+)
 from evaluation.utils.shared import (
     EvalException,
     EvalMetadata,
     EvalOutput,
     assert_and_raise,
     codeact_user_response,
+    get_metrics,
     is_fatal_evaluation_error,
     make_metadata,
     prepare_dataset,
@@ -40,9 +44,10 @@ from openhands.utils.async_utils import call_async_from_sync
 from openhands.utils.shutdown_listener import sleep_if_should_continue
 
 USE_HINT_TEXT = os.environ.get('USE_HINT_TEXT', 'false').lower() == 'true'
-USE_INSTANCE_IMAGE = os.environ.get('USE_INSTANCE_IMAGE', 'false').lower() == 'true'
+USE_INSTANCE_IMAGE = os.environ.get('USE_INSTANCE_IMAGE', 'true').lower() == 'true'
 RUN_WITH_BROWSING = os.environ.get('RUN_WITH_BROWSING', 'false').lower() == 'true'
 
+
 AGENT_CLS_TO_FAKE_USER_RESPONSE_FN = {
     'CodeActAgent': codeact_user_response,
 }
@@ -62,11 +67,11 @@ def get_instruction(instance: pd.Series, metadata: EvalMetadata):
         '<uploaded_files>\n'
         f'/workspace/{workspace_dir_name}\n'
         '</uploaded_files>\n'
-        f"I've uploaded a python code repository in the directory {workspace_dir_name}. Consider the following PR description:\n\n"
-        f'<pr_description>\n'
+        f"I've uploaded a python code repository in the directory {workspace_dir_name}. Consider the following issue description:\n\n"
+        f'<issue_description>\n'
         f'{instance.problem_statement}\n'
-        '</pr_description>\n\n'
-        'Can you help me implement the necessary changes to the repository so that the requirements specified in the <pr_description> are met?\n'
+        '</issue_description>\n\n'
+        'Can you help me implement the necessary changes to the repository so that the requirements specified in the <issue_description> are met?\n'
         "I've already taken care of all changes to any of the test files described in the <pr_description>. This means you DON'T have to modify the testing logic or any of the tests in any way!\n"
         'Your task is to make the minimal changes to non-tests files in the /workspace directory to ensure the <pr_description> is satisfied.\n'
         'Follow these steps to resolve the issue:\n'
@@ -134,6 +139,12 @@ def get_config(
             remote_runtime_api_url=os.environ.get('SANDBOX_REMOTE_RUNTIME_API_URL'),
             keep_runtime_alive=False,
             remote_runtime_init_timeout=3600,
+            remote_runtime_api_timeout=120,
+            remote_runtime_resource_factor=get_instance_resource_factor(
+                dataset_name=metadata.dataset,
+                instance_id=instance['instance_id'],
+            ),
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -148,6 +159,8 @@ def get_config(
         codeact_enable_jupyter=False,
         codeact_enable_browsing=RUN_WITH_BROWSING,
         codeact_enable_llm_editor=False,
+        condenser=metadata.condenser_config,
+        enable_prompt_extensions=False,
     )
     config.set_agent_config(agent_config)
     return config
@@ -171,7 +184,7 @@ def initialize_runtime(
     action = CmdRunAction(
         command=f"""echo 'export SWE_INSTANCE_ID={instance['instance_id']}' >> ~/.bashrc && echo 'export PIP_CACHE_DIR=~/.cache/pip' >> ~/.bashrc && echo "alias git='git --no-pager'" >> ~/.bashrc"""
     )
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -180,7 +193,7 @@ def initialize_runtime(
     )
 
     action = CmdRunAction(command="""export USER=$(whoami); echo USER=${USER} """)
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -192,7 +205,7 @@ def initialize_runtime(
 
         # inject the instance info
         action = CmdRunAction(command='mkdir -p /swe_util/eval_data/instances')
-        action.timeout = 600
+        action.set_hard_timeout(600)
         logger.info(action, extra={'msg_type': 'ACTION'})
         obs = runtime.run_action(action)
         logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -221,14 +234,14 @@ def initialize_runtime(
             '/swe_util/',
         )
         action = CmdRunAction(command='cat ~/.bashrc')
-        action.timeout = 600
+        action.set_hard_timeout(600)
         logger.info(action, extra={'msg_type': 'ACTION'})
         obs = runtime.run_action(action)
         logger.info(obs, extra={'msg_type': 'OBSERVATION'})
         assert_and_raise(obs.exit_code == 0, f'Failed to cat ~/.bashrc: {str(obs)}')
 
         action = CmdRunAction(command='source ~/.bashrc')
-        action.timeout = 600
+        action.set_hard_timeout(600)
         logger.info(action, extra={'msg_type': 'ACTION'})
         obs = runtime.run_action(action)
         logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -237,7 +250,7 @@ def initialize_runtime(
         assert_and_raise(obs.exit_code == 0, f'Failed to source ~/.bashrc: {str(obs)}')
 
         action = CmdRunAction(command='source /swe_util/instance_swe_entry.sh')
-        action.timeout = 3600
+        action.set_hard_timeout(600)
         logger.info(action, extra={'msg_type': 'ACTION'})
         obs = runtime.run_action(action)
         logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -247,7 +260,7 @@ def initialize_runtime(
         )
     else:
         action = CmdRunAction(command='source /swe_util/swe_entry.sh')
-        action.timeout = 1800
+        action.set_hard_timeout(1800)
         logger.info(action, extra={'msg_type': 'ACTION'})
         obs = runtime.run_action(action)
         logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -257,7 +270,7 @@ def initialize_runtime(
         )
 
     action = CmdRunAction(command=f'cd /workspace/{workspace_dir_name}')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -267,7 +280,7 @@ def initialize_runtime(
     )
 
     action = CmdRunAction(command='git reset --hard')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -276,14 +289,14 @@ def initialize_runtime(
     action = CmdRunAction(
         command='for remote_name in $(git remote); do git remote remove "${remote_name}"; done'
     )
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
     assert_and_raise(obs.exit_code == 0, f'Failed to remove git remotes: {str(obs)}')
 
     action = CmdRunAction(command='which python')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -314,7 +327,7 @@ def complete_runtime(
     workspace_dir_name = _get_swebench_workspace_dir_name(instance)
 
     action = CmdRunAction(command=f'cd /workspace/{workspace_dir_name}')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -324,7 +337,7 @@ def complete_runtime(
     )
 
     action = CmdRunAction(command='git config --global core.pager ""')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -334,7 +347,7 @@ def complete_runtime(
     )
 
     action = CmdRunAction(command='git add -A')
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -349,7 +362,7 @@ def complete_runtime(
         action = CmdRunAction(
             command=f'git diff --no-color --cached {instance["base_commit"]}'
         )
-        action.timeout = 600 + 100 * n_retries
+        action.set_hard_timeout(max(300 + 100 * n_retries, 600))
         logger.info(action, extra={'msg_type': 'ACTION'})
         obs = runtime.run_action(action)
         logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -397,7 +410,7 @@ def process_instance(
             8,
         )
         logger.warning(
-            f'This is the second attempt for instance {instance.instance_id}, setting resource factor to {config.sandbox.remote_runtime_resource_factor}'
+            f'This is the {runtime_failure_count + 1}th attempt for instance {instance.instance_id}, setting resource factor to {config.sandbox.remote_runtime_resource_factor}'
         )
     runtime = create_runtime(config)
     call_async_from_sync(runtime.connect)
@@ -448,7 +461,7 @@ def process_instance(
 
     # NOTE: this is NO LONGER the event stream, but an agent history that includes delegate agent's events
     histories = [event_to_dict(event) for event in state.history]
-    metrics = state.metrics.get() if state.metrics else None
+    metrics = get_metrics(state)
 
     # Save the output
     output = EvalOutput(
@@ -477,6 +490,10 @@ def filter_dataset(dataset: pd.DataFrame, filter_column: str) -> pd.DataFrame:
                 subset = dataset[dataset[filter_column].isin(selected_ids)]
                 logger.info(f'Retained {subset.shape[0]} tasks after filtering')
                 return subset
+    skip_ids = os.environ.get('SKIP_IDS', '').split(',')
+    if len(skip_ids) > 0:
+        logger.info(f'Filtering {len(skip_ids)} tasks from "SKIP_IDS"...')
+        return dataset[~dataset[filter_column].isin(skip_ids)]
     return dataset
 
 
@@ -499,8 +516,10 @@ if __name__ == '__main__':
     # NOTE: It is preferable to load datasets from huggingface datasets and perform post-processing
     # so we don't need to manage file uploading to OpenHands's repo
     dataset = load_dataset(args.dataset, split=args.split)
-    logger.info(f'Loaded dataset {args.dataset} with split {args.split}')
     swe_bench_tests = filter_dataset(dataset.to_pandas(), 'instance_id')
+    logger.info(
+        f'Loaded dataset {args.dataset} with split {args.split}: {len(swe_bench_tests)} tasks'
+    )
 
     llm_config = None
     if args.llm_config:
@@ -529,6 +548,7 @@ if __name__ == '__main__':
     )
 
     output_file = os.path.join(metadata.eval_output_dir, 'output.jsonl')
+    print(f'### OUTPUT FILE: {output_file} ###')
     instances = prepare_dataset(swe_bench_tests, output_file, args.eval_n_limit)
 
     if len(instances) > 0 and not isinstance(

evaluation/benchmarks/swe_bench/scripts/eval/combine_final_completions.py

@@ -0,0 +1,69 @@
+import argparse
+import gzip
+import json
+import os
+from glob import glob
+
+from tqdm import tqdm
+
+tqdm.pandas()
+
+
+# Load trajectories for resolved instances
+def load_completions(output_dir: str, instance_id: str):
+    glob_path = os.path.join(output_dir, 'llm_completions', instance_id, '*.json')
+    files = sorted(glob(glob_path))  # this is ascending order
+    # pick the last file (last turn)
+    try:
+        file_path = files[-1]
+    except IndexError:
+        # print(f'No files found for instance {instance_id}: files={files}')
+        return None
+    with open(file_path, 'r') as f:
+        result = json.load(f)
+    # create messages
+    messages = result['messages']
+    messages.append(result['response']['choices'][0]['message'])
+    tools = result['kwargs']['tools']
+    return {
+        'messages': messages,
+        'tools': tools,
+    }
+
+
+parser = argparse.ArgumentParser()
+parser.add_argument('jsonl_path', type=str)
+args = parser.parse_args()
+
+output_dir = os.path.dirname(args.jsonl_path)
+output_path = os.path.join(output_dir, 'output.with_completions.jsonl.gz')
+
+# Check if output would be different from input
+needs_update = False
+with open(args.jsonl_path, 'r') as f_in:
+    for line in tqdm(f_in, desc='Checking for changes'):
+        data = json.loads(line)
+        new_completions = load_completions(output_dir, data['instance_id'])
+        current_completions = data.get('raw_completions')
+        if current_completions != new_completions:
+            needs_update = True
+            break
+
+if not needs_update:
+    print('No updates required. Skipping file update.')
+    exit(0)
+
+if os.path.exists(output_path):
+    print(f'Output file already exists at {output_path}, overwriting? (y/n)')
+    if input() != 'y':
+        print('Exiting...')
+        exit(0)
+
+# Process line by line
+with open(args.jsonl_path, 'r') as f_in, gzip.open(output_path, 'wt') as f_out:
+    for line in tqdm(f_in):
+        data = json.loads(line)
+        data['raw_completions'] = load_completions(output_dir, data['instance_id'])
+        f_out.write(json.dumps(data) + '\n')
+
+print(f'Saved compressed output to {output_path}')

evaluation/benchmarks/swe_bench/scripts/eval/compare_outputs.py

@@ -15,11 +15,25 @@ parser.add_argument(
     action='store_true',
     help='Show visualization paths for failed instances',
 )
+parser.add_argument(
+    '--only-x-instances',
+    action='store_true',
+    help='Only show instances that are ran by X',
+)
 args = parser.parse_args()
 
 df1 = pd.read_json(args.input_file_1, orient='records', lines=True)
 df2 = pd.read_json(args.input_file_2, orient='records', lines=True)
 
+if args.only_x_instances:
+    instance_ids_1 = set(df1['instance_id'].tolist())
+    print(
+        f'Before removing instances not in X={args.input_file_1}: Y={df2.shape[0]} instances'
+    )
+    df2 = df2[df2['instance_id'].isin(instance_ids_1)]
+    print(
+        f'After removing instances not in X={args.input_file_1}: Y={df2.shape[0]} instances'
+    )
 
 # Get the intersection of the instance_ids
 df = pd.merge(df1, df2, on='instance_id', how='inner')
@@ -86,7 +100,7 @@ repo_diffs = []
 for repo in all_repos:
     x_count = len(x_only_by_repo.get(repo, []))
     y_count = len(y_only_by_repo.get(repo, []))
-    diff = abs(x_count - y_count)
+    diff = y_count - x_count
     repo_diffs.append((repo, diff))
 
 # Sort by diff (descending) and then by repo name
@@ -106,7 +120,13 @@ for repo, diff in repo_diffs:
     repo_color = 'red' if is_significant else 'yellow'
 
     print(f"\n{colored(repo, repo_color, attrs=['bold'])}:")
-    print(colored(f'Difference: {diff} instances!', repo_color, attrs=['bold']))
+    print(
+        colored(
+            f'Difference: {diff} instances! (Larger diff = Y better)',
+            repo_color,
+            attrs=['bold'],
+        )
+    )
     print(colored(f'X resolved but Y failed: ({len(x_instances)} instances)', 'green'))
     if x_instances:
         print('  ' + str(x_instances))

evaluation/benchmarks/swe_bench/scripts/eval/convert_oh_output_to_md.py

@@ -4,6 +4,7 @@
 import argparse
 import json
 import os
+from glob import glob
 
 import pandas as pd
 from tqdm import tqdm
@@ -20,6 +21,7 @@ output_md_folder = args.oh_output_file.replace('.jsonl', '.viz')
 print(f'Converting {args.oh_output_file} to markdown files in {output_md_folder}')
 
 oh_format = pd.read_json(args.oh_output_file, orient='records', lines=True)
+output_dir = os.path.dirname(args.oh_output_file)
 
 swebench_eval_file = args.oh_output_file.replace('.jsonl', '.swebench_eval.jsonl')
 if os.path.exists(swebench_eval_file):
@@ -57,22 +59,172 @@ def convert_history_to_str(history):
     return ret
 
 
+# Load trajectories for resolved instances
+def load_completions(instance_id: str):
+    global output_dir
+    glob_path = os.path.join(output_dir, 'llm_completions', instance_id, '*.json')
+    files = sorted(glob(glob_path))  # this is ascending order
+    # pick the last file (last turn)
+    try:
+        file_path = files[-1]
+    except IndexError:
+        # print(f'No files found for instance {instance_id}: files={files}')
+        return None
+    with open(file_path, 'r') as f:
+        result = json.load(f)
+    # create messages
+    messages = result['messages']
+    messages.append(result['response']['choices'][0]['message'])
+    tools = result['kwargs']['tools']
+    return {
+        'messages': messages,
+        'tools': tools,
+    }
+
+
+def _convert_content(content) -> str:
+    ret = ''
+    if isinstance(content, list):
+        for item in content:
+            assert item['type'] == 'text', 'Only text is supported for now'
+            ret += f'{item["text"]}\n'
+    else:
+        assert isinstance(content, str), 'Only str is supported for now'
+        ret = content
+    return ret
+
+
+def convert_tool_call_to_string(tool_call: dict) -> str:
+    """Convert tool call to content in string format."""
+    if 'function' not in tool_call:
+        raise ValueError("Tool call must contain 'function' key.")
+    if 'id' not in tool_call:
+        raise ValueError("Tool call must contain 'id' key.")
+    if 'type' not in tool_call:
+        raise ValueError("Tool call must contain 'type' key.")
+    if tool_call['type'] != 'function':
+        raise ValueError("Tool call type must be 'function'.")
+
+    ret = f"<function={tool_call['function']['name']}>\n"
+    try:
+        args = json.loads(tool_call['function']['arguments'])
+    except json.JSONDecodeError as e:
+        raise ValueError(
+            f"Failed to parse arguments as JSON. Arguments: {tool_call['function']['arguments']}"
+        ) from e
+    for param_name, param_value in args.items():
+        is_multiline = isinstance(param_value, str) and '\n' in param_value
+        ret += f'<parameter={param_name}>'
+        if is_multiline:
+            ret += '\n'
+        ret += f'{param_value}'
+        if is_multiline:
+            ret += '\n'
+        ret += '</parameter>\n'
+    ret += '</function>'
+    return ret
+
+
+def format_traj(traj, first_n_turns=None, last_n_turns=None) -> str:
+    output = ''
+    system_message = None
+
+    # Handle system message if present
+    if traj[0]['role'] == 'system':
+        system_message = traj[0]
+        traj = traj[1:]
+        content = _convert_content(system_message['content'])
+        output += "*** System Message that describes the assistant's behavior ***\n"
+        output += f'{content}\n'
+
+    # Merge consecutive user messages first
+    merged_traj = []
+    current_messages = []
+
+    n_turns = len(traj)
+    for i, message in enumerate(traj):
+        # Skip this message if...
+        if (
+            # Case 1: first_n_turns specified and we're past it
+            (first_n_turns is not None and i >= first_n_turns and last_n_turns is None)
+            or
+            # Case 2: last_n_turns specified and we're before it
+            (
+                last_n_turns is not None
+                and i < n_turns - last_n_turns
+                and first_n_turns is None
+            )
+            or
+            # Case 3: both specified and we're in the middle section
+            (
+                first_n_turns is not None
+                and last_n_turns is not None
+                and i >= first_n_turns
+                and i < n_turns - last_n_turns
+            )
+        ):
+            continue
+
+        if message['role'] == 'user':
+            current_messages.append(message)
+        else:
+            if current_messages:
+                # Merge all accumulated user messages into one
+                merged_content = '\n'.join(
+                    _convert_content(msg['content']) for msg in current_messages
+                )
+                merged_traj.append({'role': 'user', 'content': merged_content})
+                current_messages = []
+            merged_traj.append(message)
+
+    # Don't forget to handle any remaining user messages
+    if current_messages:
+        merged_content = '\n'.join(
+            _convert_content(msg['content']) for msg in current_messages
+        )
+        merged_traj.append({'role': 'user', 'content': merged_content})
+
+    # Now process the merged trajectory
+    for i, message in enumerate(merged_traj):
+        role, content = message['role'], message['content']
+        content = _convert_content(content) if isinstance(content, list) else content
+        turn_id = i // 2 + 1
+        output += '-' * 100 + '\n'
+        output += f'*** Turn {turn_id} - {role.upper() if role != "tool" else "TOOL EXECUTION RESULT"} ***\n'
+
+        if role == 'user':
+            output += f'{content}\n'
+        elif role == 'tool':
+            output += f'{content}\n'
+        elif role == 'assistant':
+            output += f'{content}\n'
+            if (
+                'tool_calls' in message
+                and message['tool_calls'] is not None
+                and len(message['tool_calls']) > 0
+            ):
+                for toolcall_id, tool_call in enumerate(message['tool_calls']):
+                    output += f'### Tool Call {toolcall_id}\n'
+                    output += f'{convert_tool_call_to_string(tool_call)}\n'
+        else:
+            raise ValueError(f'Unexpected role: {role}')
+
+    output += '-' * 100 + '\n'
+    return output
+
+
 def write_row_to_md_file(row, instance_id_to_test_result):
     if 'git_patch' in row:
         model_patch = row['git_patch']
     elif 'test_result' in row and 'git_patch' in row['test_result']:
         model_patch = row['test_result']['git_patch']
     else:
-        raise ValueError(f'Row {row} does not have a git_patch')
+        print(f'Row {row} does not have a git_patch')
+        return
 
     test_output = None
-    if row['instance_id'] in instance_id_to_test_result:
-        report = instance_id_to_test_result[row['instance_id']].get('report', {})
-        resolved = report.get('resolved', False)
-        test_output = instance_id_to_test_result[row['instance_id']].get(
-            'test_output', None
-        )
-    elif 'report' in row and row['report'] is not None:
+    # Use result from output.jsonl FIRST if available.
+    if 'report' in row and row['report'] is not None:
         if not isinstance(row['report'], dict):
             resolved = None
             print(
@@ -80,6 +232,12 @@ def write_row_to_md_file(row, instance_id_to_test_result):
             )
         else:
             resolved = row['report'].get('resolved', False)
+    elif row['instance_id'] in instance_id_to_test_result:
+        report = instance_id_to_test_result[row['instance_id']].get('report', {})
+        resolved = report.get('resolved', False)
+        test_output = instance_id_to_test_result[row['instance_id']].get(
+            'test_output', None
+        )
     else:
         resolved = None
 
@@ -88,6 +246,8 @@ def write_row_to_md_file(row, instance_id_to_test_result):
     os.makedirs(output_md_folder, exist_ok=True)
     filepath = os.path.join(output_md_folder, filename)
 
+    completions = load_completions(instance_id)
+
     with open(filepath, 'w') as f:
         f.write(f'# {instance_id} (resolved: {resolved})\n')
 
@@ -97,7 +257,12 @@ def write_row_to_md_file(row, instance_id_to_test_result):
         f.write(json.dumps(row['metadata'], indent=2))
         f.write('\n```\n')
 
-        # Trajectory
+        # Completion
+        if completions is not None:
+            f.write('## Completion\n')
+            traj = completions['messages']
+            f.write(format_traj(traj))
+
         f.write('## History\n')
         f.write(convert_history_to_str(row['history']))
 

evaluation/benchmarks/swe_bench/scripts/eval/convert_oh_output_to_swe_json.py

@@ -22,7 +22,8 @@ def convert_row_to_swebench_format(row):
     elif 'test_result' in row and 'git_patch' in row['test_result']:
         model_patch = row['test_result']['git_patch']
     else:
-        raise ValueError(f'Row {row} does not have a git_patch')
+        print(f'WARNING: Row {row} does not have a git_patch')
+        model_patch = ''
 
     return {
         'instance_id': row['instance_id'],

evaluation/benchmarks/swe_bench/scripts/eval/update_output_with_eval.py

@@ -3,7 +3,7 @@ import json
 import os
 from collections import defaultdict
 
-import pandas as pd
+from tqdm import tqdm
 
 parser = argparse.ArgumentParser()
 parser.add_argument('input_file', type=str)
@@ -11,8 +11,7 @@ args = parser.parse_args()
 
 dirname = os.path.dirname(args.input_file)
 
-df = pd.read_json(args.input_file, lines=True)
-
+# Initialize counters and data structures
 instance_id_to_status = defaultdict(
     lambda: {
         'empty_generation': False,
@@ -23,15 +22,7 @@ instance_id_to_status = defaultdict(
     }
 )
 
-
-# Apply the status to the dataframe
-def apply_report(row):
-    instance_id = row['instance_id']
-    if instance_id in instance_id_to_status:
-        return dict(instance_id_to_status[instance_id])
-    return row.get('report', {})
-
-
+# Process official report if it exists
 swebench_official_report_json = os.path.join(dirname, 'report.json')
 openhands_remote_report_jsonl = args.input_file.replace(
     '.jsonl', '.swebench_eval.jsonl'
@@ -90,113 +81,159 @@ if os.path.exists(swebench_official_report_json):
             f'- [{instance_id}](./eval_outputs/{instance_id}/run_instance.log)\n'
         )
 
-    df['report'] = df.apply(apply_report, axis=1)
-
     with open(output_md_filepath, 'w') as f:
         f.write(output_md)
 
 elif os.path.exists(openhands_remote_report_jsonl):
     output_md_filepath = args.input_file.replace('.jsonl', '.swebench_eval.md')
 
-    df_eval = pd.read_json(openhands_remote_report_jsonl, lines=True, orient='records')
+    # First pass: Read eval report and count instances
+    instance_ids = set()
+    eval_instance_ids = set()
 
-    assert len(df['instance_id'].unique()) == len(
-        df
-    ), 'There are duplicate instance ids in the original output which is not allowed'
-    assert len(df_eval['instance_id'].unique()) == len(
-        df_eval
-    ), 'There are duplicate instance ids in the eval report which is not allowed'
+    # Count instances in original file
+    n_instances = 0
+    with open(args.input_file, 'r') as f:
+        for line in tqdm(f, desc='Counting instances in original file'):
+            data = json.loads(line)
+            instance_ids.add(data['instance_id'])
+            n_instances += 1
+    print(f'Total instances in original file: {n_instances}')
 
-    for _, row in df_eval.iterrows():
-        instance_id_to_status[row['instance_id']] = row['test_result']['report']
-    df['report'] = df.apply(apply_report, axis=1)
+    # Process eval report
+    n_eval_instances = 0
+    with open(openhands_remote_report_jsonl, 'r') as f:
+        for line in tqdm(f, desc='Processing eval report'):
+            data = json.loads(line)
+            instance_id = data['instance_id']
+            eval_instance_ids.add(instance_id)
+            n_eval_instances += 1
+            instance_id_to_status[instance_id] = data['test_result']['report']
+    print(f'Total instances in eval report: {n_eval_instances}')
 
-    report_is_dict = df['report'].apply(lambda x: isinstance(x, dict))
-    if not report_is_dict.all():
-        print(df[~report_is_dict])
-        raise ValueError(f'Report is not a dict, but a {type(row["report"])}')
+    # Verify no duplicates
+    assert (
+        len(instance_ids) == n_instances
+    ), 'Duplicate instance ids found in original output'
+    assert (
+        len(eval_instance_ids) == n_eval_instances
+    ), 'Duplicate instance ids found in eval report'
 
-    _n_instances = len(df)
-    _n_resolved = len(df[df['report'].apply(lambda x: x.get('resolved', False))])
-    _n_unresolved = _n_instances - _n_resolved
-    _n_empty_patch = len(
-        df[df['report'].apply(lambda x: x.get('empty_generation', False))]
-    )
-    _n_error = len(df[df['report'].apply(lambda x: x.get('error_eval', False))])
+    # Initialize counters
+    stats = {'total': len(instance_ids), 'resolved': 0, 'empty_patch': 0, 'error': 0}
+
+    # Collect instance IDs by category
+    resolved_ids = []
+    unresolved_ids = []
+    error_ids = []
+    empty_patch_ids = []
+    timeout_ids = []
+
+    # Process original file and categorize instances
+    with open(args.input_file, 'r') as f:
+        for line in f:
+            data = json.loads(line)
+            instance_id = data['instance_id']
+            report = instance_id_to_status[instance_id]
+
+            if report.get('resolved', False):
+                stats['resolved'] += 1
+                resolved_ids.append(instance_id)
+            else:
+                unresolved_ids.append(instance_id)
+
+            if report.get('empty_generation', False):
+                stats['empty_patch'] += 1
+                empty_patch_ids.append(instance_id)
+            if report.get('error_eval', False):
+                stats['error'] += 1
+                error_ids.append(instance_id)
+            if report.get('test_timeout', False):
+                timeout_ids.append(instance_id)
+
+    # Generate markdown report
+    def _instance_id_to_log_path(instance_id):
+        path = f"{args.input_file.replace('.jsonl', '.swebench_eval.logs')}/instance_{instance_id}.log"
+        return os.path.relpath(path, start=dirname)
+
+    # ... rest of markdown generation code remains the same ...
     output_md = (
         '# SWE-bench Report\n'
         'This folder contains the evaluation results of the SWE-bench using the [official evaluation docker containerization](https://github.com/princeton-nlp/SWE-bench/blob/main/docs/20240627_docker/README.md#choosing-the-right-cache_level).\n\n'
         '## Summary\n'
-        f'- submitted instances: {_n_instances}\n'
-        f'- empty patch instances: {_n_empty_patch}\n'
-        f'- resolved instances: {_n_resolved}\n'
-        f'- unresolved instances: {_n_unresolved}\n'
-        f'- error instances: {_n_error}\n'
+        f'- submitted instances: {stats["total"]}\n'
+        f'- empty patch instances: {stats["empty_patch"]}\n'
+        f'- resolved instances: {stats["resolved"]}\n'
+        f'- unresolved instances: {len(unresolved_ids)}\n'
+        f'- error instances: {stats["error"]}\n'
     )
 
-    def _instance_id_to_log_path(instance_id):
-        path = f"{args.input_file.replace('.jsonl', '.swebench_eval.logs')}/instance_{instance_id}.log"
-        # make it relative path
-        path = os.path.relpath(path, start=dirname)
-        return path
-
     output_md += '\n## Resolved Instances\n'
     # instance_id to status
-    for instance_id in sorted(
-        df[df['report'].apply(lambda x: x.get('resolved', False))][
-            'instance_id'
-        ].unique()
-    ):
+    for instance_id in resolved_ids:
         instance_id_to_status[instance_id]['resolved'] = True
         output_md += f'- [{instance_id}]({_instance_id_to_log_path(instance_id)})\n'
 
     output_md += '\n## Unresolved Instances\n'
-    for instance_id in sorted(
-        df[~df['report'].apply(lambda x: x.get('resolved', False))][
-            'instance_id'
-        ].unique()
-    ):
+    for instance_id in unresolved_ids:
         output_md += f'- [{instance_id}]({_instance_id_to_log_path(instance_id)})\n'
 
     output_md += '\n## Error Instances\n'
-    for instance_id in sorted(
-        df[df['report'].apply(lambda x: x.get('error_eval', False))][
-            'instance_id'
-        ].unique()
-    ):
+    for instance_id in error_ids:
         instance_id_to_status[instance_id]['error_eval'] = True
         output_md += f'- [{instance_id}]({_instance_id_to_log_path(instance_id)})\n'
 
     output_md += '\n## Empty Patch Instances\n'
-    for instance_id in sorted(
-        df[df['report'].apply(lambda x: x.get('empty_generation', False))][
-            'instance_id'
-        ].unique()
-    ):
+    for instance_id in empty_patch_ids:
         instance_id_to_status[instance_id]['empty_generation'] = True
         output_md += f'- [{instance_id}]({_instance_id_to_log_path(instance_id)})\n'
 
     output_md += '\n## Incomplete Instances\n'
-    for instance_id in sorted(
-        df[df['report'].apply(lambda x: x.get('test_timeout', False))][
-            'instance_id'
-        ].unique()
-    ):
+    for instance_id in timeout_ids:
         output_md += f'- [{instance_id}]({_instance_id_to_log_path(instance_id)})\n'
+
     with open(output_md_filepath, 'w') as f:
         f.write(output_md)
+
 else:
     print(
         f'No report file found: Both {swebench_official_report_json} and {openhands_remote_report_jsonl} do not exist.'
     )
     exit()
 
+# Before backup and update, check if any changes would be made
+needs_update = False
+with open(args.input_file, 'r') as infile:
+    for line in tqdm(infile, desc='Checking for changes'):
+        data = json.loads(line)
+        instance_id = data['instance_id']
+        current_report = data.get('report', {})
+        new_report = instance_id_to_status[
+            instance_id
+        ]  # if no report, it's not resolved
+        if current_report != new_report:
+            needs_update = True
+            break
+
+if not needs_update:
+    print('No updates detected. Skipping file update.')
+    exit()
+
+# Backup and update the original file row by row
 if os.path.exists(args.input_file + '.bak'):
     conf = input('Existing backup file found. Do you want to overwrite it? (y/n)')
     if conf != 'y':
         exit()
     os.remove(args.input_file + '.bak')
 
-# backup the original file
 os.rename(args.input_file, args.input_file + '.bak')
-df.to_json(args.input_file, orient='records', lines=True)
+
+# Process and write file row by row
+with open(args.input_file + '.bak', 'r') as infile, open(
+    args.input_file, 'w'
+) as outfile:
+    for line in tqdm(infile, desc='Updating output file'):
+        data = json.loads(line)
+        instance_id = data['instance_id']
+        data['report'] = instance_id_to_status[instance_id]
+        outfile.write(json.dumps(data) + '\n')

evaluation/benchmarks/swe_bench/scripts/eval_infer.sh

@@ -76,7 +76,7 @@ echo "Running SWE-bench evaluation"
 echo "=============================================================="
 
 RUN_ID=$(date +"%Y%m%d_%H%M%S")
-N_PROCESS=16
+N_PROCESS=4
 
 if [ -z "$INSTANCE_ID" ]; then
     echo "Running SWE-bench evaluation on the whole input file..."
@@ -87,7 +87,7 @@ if [ -z "$INSTANCE_ID" ]; then
         --dataset_name "$DATASET_NAME" \
         --split "$SPLIT" \
         --predictions_path $SWEBENCH_FORMAT_JSONL \
-        --timeout 1800 \
+        --timeout 3600 \
         --cache_level instance \
         --max_workers $N_PROCESS \
         --run_id $RUN_ID
@@ -133,7 +133,7 @@ else
         --dataset_name "$DATASET_NAME" \
         --split "$SPLIT" \
         --predictions_path $SWEBENCH_FORMAT_JSONL \
-        --timeout 1800 \
+        --timeout 3600 \
         --instance_ids $INSTANCE_ID \
         --cache_level instance \
         --max_workers $N_PROCESS \

evaluation/benchmarks/swe_bench/scripts/run_infer.sh

@@ -108,7 +108,14 @@ if [ -z "$N_RUNS" ]; then
   echo "N_RUNS not specified, use default $N_RUNS"
 fi
 
+# Skip runs if the run number is in the SKIP_RUNS list
+# read from env variable SKIP_RUNS as a comma separated list of run numbers
+SKIP_RUNS=(${SKIP_RUNS//,/ })
 for i in $(seq 1 $N_RUNS); do
+  if [[ " ${SKIP_RUNS[@]} " =~ " $i " ]]; then
+    echo "Skipping run $i"
+    continue
+  fi
   current_eval_note="$EVAL_NOTE-run_$i"
   echo "EVAL_NOTE: $current_eval_note"
   run_eval $current_eval_note

evaluation/benchmarks/the_agent_company/README.md

@@ -17,11 +17,14 @@ When the `run_infer.sh` script is started, it will automatically pull all task i
 
 ```bash
 ./evaluation/benchmarks/the_agent_company/scripts/run_infer.sh \
-  --agent-llm-config <agent-llm-config>  \
-  --env-llm-config <env-llm-config> \
-  --outputs-path <outputs-path> \
-  --server-hostname <server-hostname> \
-  --version <version>
+  --agent-llm-config <agent-llm-config, default to 'agent'>  \
+  --env-llm-config <env-llm-config, default to 'env'> \
+  --outputs-path <outputs-path, default to outputs> \
+  --server-hostname <server-hostname, default to localhost> \
+  --version <version, default to 1.0.0> \
+  --start-percentile <integer from 0 to 99, default to 0> \
+  --end-percentile <integer from 1 to 100, default to 100>
+
 
 # Example
 ./evaluation/benchmarks/the_agent_company/scripts/run_infer.sh \
@@ -29,7 +32,9 @@ When the `run_infer.sh` script is started, it will automatically pull all task i
   --env-llm-config claude-3-5-sonnet-20240620 \
   --outputs-path outputs \
   --server-hostname localhost \
-  --version 1.0.0
+  --version 1.0.0 \
+  --start-percentile 10 \
+  --end-percentile 20
 ```
 
 - `agent-llm-config`: the config name for the agent LLM. This should match the config name in config.toml. This is the LLM used by the agent (e.g. CodeActAgent).
@@ -37,7 +42,11 @@ When the `run_infer.sh` script is started, it will automatically pull all task i
 - `outputs-path`: the path to save trajectories and evaluation results.
 - `server-hostname`: the hostname of the server that hosts all the web services. It could be localhost if you are running the evaluation and services on the same machine. If the services are hosted on a remote machine, you must use the hostname of the remote machine rather than IP address.
 - `version`: the version of the task images to use. Currently, the only supported version is 1.0.0.
+- `start-percentile`: the start percentile of the task split, must be an integer between 0 to 99.
+- `end-percentile`: the end percentile of the task split, must be an integer between 1 to 100 and larger than start-percentile.
 
-The script is idempotent. If you run it again, it will resume from the last checkpoint. It would usually take a few days to finish evaluation.
+The script is idempotent. If you run it again, it will resume from the last checkpoint. It would usually take 2 days to finish evaluation if you run the whole task set.
+To speed up evaluation, you can use `start-percentile` and `end-percentile` to split the tasks for higher parallelism,
+provided concurrent runs are **targeting different servers**.
 
 Note: the script will automatically skip a task if it encounters an error. This usually happens when the OpenHands runtime dies due to some unexpected errors. This means even if the script finishes, it might not have evaluated all tasks. You can manually resume the evaluation by running the script again.

evaluation/benchmarks/the_agent_company/browsing.py

@@ -262,7 +262,7 @@ def pre_login(
             instruction = action.to_instruction()
 
             browser_action = BrowseInteractiveAction(browser_actions=instruction)
-            browser_action.timeout = 10000
+            browser_action.set_hard_timeout(10000)
             logger.info(browser_action, extra={'msg_type': 'ACTION'})
             obs: BrowserOutputObservation = runtime.run_action(browser_action)
             logger.debug(obs, extra={'msg_type': 'OBSERVATION'})

evaluation/benchmarks/the_agent_company/run_infer.py

@@ -18,9 +18,11 @@ from openhands.core.config import (
     AppConfig,
     LLMConfig,
     SandboxConfig,
+    get_agent_config_arg,
     get_llm_config_arg,
     get_parser,
 )
+from openhands.core.config.agent_config import AgentConfig
 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
@@ -34,12 +36,13 @@ def get_config(
     task_short_name: str,
     mount_path_on_host: str,
     llm_config: LLMConfig,
+    agent_config: AgentConfig,
 ) -> AppConfig:
     config = AppConfig(
         run_as_openhands=False,
         max_budget_per_task=4,
         max_iterations=100,
-        trajectories_path=os.path.join(
+        save_trajectory_path=os.path.join(
             mount_path_on_host, f'traj_{task_short_name}.json'
         ),
         sandbox=SandboxConfig(
@@ -50,6 +53,7 @@ def get_config(
             # large enough timeout, since some testcases take very long to run
             timeout=300,
             api_key=os.environ.get('ALLHANDS_API_KEY', None),
+            remote_runtime_enable_retries=True,
         ),
         # we mount trajectories path so that trajectories, generated by OpenHands
         # controller, can be accessible to the evaluator file in the runtime container
@@ -57,6 +61,14 @@ def get_config(
         workspace_mount_path_in_sandbox='/outputs',
     )
     config.set_llm_config(llm_config)
+    if agent_config:
+        config.set_agent_config(agent_config)
+    else:
+        logger.info('Agent config not provided, using default settings')
+        agent_config = AgentConfig(
+            enable_prompt_extensions=False,
+        )
+        config.set_agent_config(agent_config)
     return config
 
 
@@ -80,13 +92,13 @@ def load_dependencies(runtime: Runtime) -> List[str]:
 def init_task_env(runtime: Runtime, hostname: str, env_llm_config: LLMConfig):
     command = (
         f'SERVER_HOSTNAME={hostname} '
-        f'LITELLM_API_KEY={env_llm_config.api_key} '
+        f'LITELLM_API_KEY={env_llm_config.api_key.get_secret_value() if env_llm_config.api_key else None} '
         f'LITELLM_BASE_URL={env_llm_config.base_url} '
         f'LITELLM_MODEL={env_llm_config.model} '
         'bash /utils/init.sh'
     )
     action = CmdRunAction(command=command)
-    action.timeout = 900
+    action.set_hard_timeout(900)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -165,14 +177,14 @@ def run_evaluator(
     runtime: Runtime, env_llm_config: LLMConfig, trajectory_path: str, result_path: str
 ):
     command = (
-        f'LITELLM_API_KEY={env_llm_config.api_key} '
+        f'LITELLM_API_KEY={env_llm_config.api_key.get_secret_value() if env_llm_config.api_key else None} '
         f'LITELLM_BASE_URL={env_llm_config.base_url} '
         f'LITELLM_MODEL={env_llm_config.model} '
         f"DECRYPTION_KEY='theagentcompany is all you need' "  # Hardcoded Key
         f'python_default /utils/eval.py --trajectory_path {trajectory_path} --result_path {result_path}'
     )
     action = CmdRunAction(command=command)
-    action.timeout = 600
+    action.set_hard_timeout(600)
     logger.info(action, extra={'msg_type': 'ACTION'})
     obs = runtime.run_action(action)
     logger.info(obs, extra={'msg_type': 'OBSERVATION'})
@@ -214,6 +226,10 @@ if __name__ == '__main__':
     )
     args, _ = parser.parse_known_args()
 
+    agent_config: AgentConfig | None = None
+    if args.agent_config:
+        agent_config = get_agent_config_arg(args.agent_config)
+
     agent_llm_config: LLMConfig | None = None
     if args.agent_llm_config:
         agent_llm_config = get_llm_config_arg(args.agent_llm_config)
@@ -254,7 +270,7 @@ if __name__ == '__main__':
     else:
         temp_dir = tempfile.mkdtemp()
     config: AppConfig = get_config(
-        args.task_image_name, task_short_name, temp_dir, agent_llm_config
+        args.task_image_name, task_short_name, temp_dir, agent_llm_config, agent_config
     )
     runtime: Runtime = create_runtime(config)
     call_async_from_sync(runtime.connect)

evaluation/benchmarks/the_agent_company/scripts/run_infer.sh

@@ -44,6 +44,10 @@ while [[ $# -gt 0 ]]; do
             ENV_LLM_CONFIG="$2"
             shift 2
             ;;
+        --agent-config)
+            AGENT_CONFIG="$2"
+            shift 2
+            ;;
         --outputs-path)
             OUTPUTS_PATH="$2"
             shift 2
@@ -56,6 +60,14 @@ while [[ $# -gt 0 ]]; do
             VERSION="$2"
             shift 2
             ;;
+        --start-percentile)
+            START_PERCENTILE="$2"
+            shift 2
+            ;;
+        --end-percentile)
+            END_PERCENTILE="$2"
+            shift 2
+            ;;
         *)
             echo "Unknown argument: $1"
             exit 1
@@ -69,16 +81,53 @@ if [[ ! "$OUTPUTS_PATH" = /* ]]; then
     OUTPUTS_PATH="$(cd "$(dirname "$OUTPUTS_PATH")" 2>/dev/null && pwd)/$(basename "$OUTPUTS_PATH")"
 fi
 
+: "${START_PERCENTILE:=0}"  # Default to 0 percentile (first line)
+: "${END_PERCENTILE:=100}"  # Default to 100 percentile (last line)
+
+# Validate percentile ranges if provided
+if ! [[ "$START_PERCENTILE" =~ ^[0-9]+$ ]] || ! [[ "$END_PERCENTILE" =~ ^[0-9]+$ ]]; then
+    echo "Error: Percentiles must be integers"
+    exit 1
+fi
+
+if [ "$START_PERCENTILE" -ge "$END_PERCENTILE" ]; then
+    echo "Error: Start percentile must be less than end percentile"
+    exit 1
+fi
+
+if [ "$START_PERCENTILE" -lt 0 ] || [ "$END_PERCENTILE" -gt 100 ]; then
+    echo "Error: Percentiles must be between 0 and 100"
+    exit 1
+fi
+
 echo "Using agent LLM config: $AGENT_LLM_CONFIG"
 echo "Using environment LLM config: $ENV_LLM_CONFIG"
 echo "Outputs path: $OUTPUTS_PATH"
 echo "Server hostname: $SERVER_HOSTNAME"
 echo "Version: $VERSION"
+echo "Start Percentile: $START_PERCENTILE"
+echo "End Percentile: $END_PERCENTILE"
 
 echo "Downloading tasks.md..."
 rm -f tasks.md
 wget https://github.com/TheAgentCompany/TheAgentCompany/releases/download/${VERSION}/tasks.md
 
+total_lines=$(cat tasks.md | grep "ghcr.io/theagentcompany" | wc -l)
+if [ "$total_lines" -ne 175 ]; then
+    echo "Error: Expected 175 tasks in tasks.md but found $total_lines lines"
+    exit 1
+fi
+
+# Calculate line numbers based on percentiles
+start_line=$(echo "scale=0; ($total_lines * $START_PERCENTILE / 100) + 1" | bc)
+end_line=$(echo "scale=0; $total_lines * $END_PERCENTILE / 100" | bc)
+
+echo "Using tasks No. $start_line to $end_line (inclusive) out of 1-175 tasks"
+
+# Create a temporary file with just the desired range
+temp_file="tasks_${START_PERCENTILE}_${END_PERCENTILE}.md"
+sed -n "${start_line},${end_line}p" tasks.md > "$temp_file"
+
 while IFS= read -r task_image; do
     docker pull $task_image
 
@@ -95,21 +144,29 @@ while IFS= read -r task_image; do
         continue
     fi
 
-    export PYTHONPATH=evaluation/benchmarks/the_agent_company:\$PYTHONPATH && \
-        poetry run python run_infer.py \
-            --agent-llm-config "$AGENT_LLM_CONFIG" \
-            --env-llm-config "$ENV_LLM_CONFIG" \
-            --outputs-path "$OUTPUTS_PATH" \
-            --server-hostname "$SERVER_HOSTNAME" \
-            --task-image-name "$task_image"
+    # Build the Python command
+    COMMAND="poetry run python run_infer.py \
+            --agent-llm-config \"$AGENT_LLM_CONFIG\" \
+            --env-llm-config \"$ENV_LLM_CONFIG\" \
+            --outputs-path \"$OUTPUTS_PATH\" \
+            --server-hostname \"$SERVER_HOSTNAME\" \
+            --task-image-name \"$task_image\""
+
+    # Add agent-config if it's defined
+    if [ -n "$AGENT_CONFIG" ]; then
+        COMMAND="$COMMAND --agent-config $AGENT_CONFIG"
+    fi
+
+    export PYTHONPATH=evaluation/benchmarks/the_agent_company:$PYTHONPATH && \
+        eval "$COMMAND"
 
     # Prune unused images and volumes
     docker image rm "$task_image"
     docker images "ghcr.io/all-hands-ai/runtime" -q | xargs -r docker rmi -f
     docker volume prune -f
     docker system prune -f
-done < tasks.md
+done < "$temp_file"
 
-rm tasks.md
+rm tasks.md "$temp_file"
 
 echo "All evaluation completed successfully!"

evaluation/benchmarks/toolqa/run_infer.py

@@ -50,6 +50,7 @@ def get_config(
             base_container_image='python:3.12-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
+            remote_runtime_enable_retries=True,
         ),
         # do not mount workspace
         workspace_base=None,
@@ -57,7 +58,7 @@ def get_config(
     )
     config.set_llm_config(metadata.llm_config)
     agent_config = config.get_agent_config(metadata.agent_class)
-    agent_config.use_microagents = False
+    agent_config.enable_prompt_extensions = False
     return config
 
 

evaluation/benchmarks/visualwebarena/README.md

@@ -0,0 +1,50 @@
+# VisualWebArena Evaluation with OpenHands Browsing Agents
+
+This folder contains evaluation for [VisualWebArena](https://github.com/web-arena-x/visualwebarena) benchmark, powered by [BrowserGym](https://github.com/ServiceNow/BrowserGym) for easy evaluation of how well an agent capable of browsing can perform on realistic web browsing tasks.
+
+## Setup Environment and LLM Configuration
+
+Please follow instruction [here](../../README.md#setup) to setup your local development environment and LLM.
+
+## Setup VisualWebArena Environment
+
+VisualWebArena requires you to set up websites containing pre-populated content that is accessible via URL to the machine running the OpenHands agents.
+Follow [this document](https://github.com/web-arena-x/visualwebarena/blob/main/environment_docker/README.md) to set up your own VisualWebArena environment through local servers or AWS EC2 instances.
+Take note of the base URL (`$VISUALWEBARENA_BASE_URL`) of the machine where the environment is installed.
+
+## Test if your environment works
+
+Access with browser the above VisualWebArena website URLs and see if they load correctly.
+If you cannot access the website, make sure the firewall allows public access of the aforementioned ports on your server
+Check the network security policy if you are using an AWS machine.
+Follow the VisualWebArena environment setup guide carefully, and make sure the URL fields are populated with the correct base URL of your server.
+
+## Run Evaluation
+
+```bash
+export VISUALWEBARENA_BASE_URL=<YOUR_SERVER_URL_HERE>
+export OPENAI_API_KEY="yourkey" # this OpenAI API key is required for some visualWebArena validators that utilize LLMs
+export OPENAI_BASE_URL="https://api.openai.com/v1/" # base URL for OpenAI model used for VisualWebArena evaluation
+bash evaluation/benchmarks/visualwebarena/scripts/run_infer.sh llm.claude HEAD VisualBrowsingAgent
+```
+
+Results will be in `evaluation/evaluation_outputs/outputs/visualwebarena/`
+
+To calculate the success rate, run:
+
+```sh
+poetry run python evaluation/benchmarks/visualwebarena/get_success_rate.py evaluation/evaluation_outputs/outputs/visualwebarena/SOME_AGENT/EXP_NAME/output.jsonl
+```
+
+## Submit your evaluation 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 following the guide [here](https://huggingface.co/docs/hub/en/repositories-pull-requests-discussions#pull-requests-and-discussions).
+
+## VisualBrowsingAgent V1.0 result
+
+Tested on VisualBrowsingAgent V1.0
+
+VisualWebArena, 910 tasks (high cost, single run due to fixed task), max step 15. Resolve rates are:
+
+- GPT4o: 26.15%
+- Claude-3.5 Sonnet: 25.27%

evaluation/benchmarks/visualwebarena/get_success_rate.py

@@ -0,0 +1,40 @@
+import argparse
+import json
+
+import browsergym.visualwebarena  # noqa F401 register visualwebarena tasks as gym environments
+import gymnasium as gym
+
+parser = argparse.ArgumentParser(description='Calculate average reward.')
+parser.add_argument('output_path', type=str, help='path to output.jsonl')
+
+args = parser.parse_args()
+
+if __name__ == '__main__':
+    env_ids = [
+        id
+        for id in gym.envs.registry.keys()
+        if id.startswith('browsergym/visualwebarena')
+    ]
+    total_num = len(env_ids)
+    print('Total number of tasks: ', total_num)
+    total_reward = 0
+    total_cost = 0
+    actual_num = 0
+    with open(args.output_path, 'r') as f:
+        for line in f:
+            data = json.loads(line)
+            actual_num += 1
+            total_cost += data['metrics']['accumulated_cost']
+            reward = data['test_result']['reward']
+            if reward >= 0:
+                total_reward += data['test_result']['reward']
+            else:
+                actual_num -= 1
+    avg_reward = total_reward / total_num
+    print('Total reward: ', total_reward)
+    print('Success Rate: ', avg_reward)
+
+    avg_cost = total_cost / actual_num
+    print('Avg Cost: ', avg_cost)
+    print('Total Cost: ', total_cost)
+    print('Actual number of tasks finished: ', actual_num)