feat(llm): Add Claude 3.7 backend configurations (#6937)

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

.github/pull_request_template.md

@@ -1,11 +1,12 @@
-**End-user friendly description of the problem this fixes or functionality that this introduces**
-
-- [ ] Include this change in the Release Notes. If checked, you must provide an **end-user friendly** description for your change below
-
----
-**Give a summary of what the PR does, explaining any non-trivial design decisions**
+- [ ] This change is worth documenting at https://docs.all-hands.dev/
+- [ ] Include this change in the Release Notes. If checked, you **must** provide an **end-user friendly** description for your change below
 
+**End-user friendly description of the problem this fixes or functionality that this introduces.**
 
 
 ---
-**Link of any specific issues this addresses**
+**Give a summary of what the PR does, explaining any non-trivial design decisions.**
+
+
+---
+**Link of any specific issues this addresses.**

Development.md

@@ -100,7 +100,7 @@ poetry run pytest ./tests/unit/test_*.py
 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:0.25-nikolaik`
+Example: `export SANDBOX_RUNTIME_CONTAINER_IMAGE=ghcr.io/all-hands-ai/runtime:0.26-nikolaik`
 
 ## Develop inside Docker container
 

Makefile

@@ -1,4 +1,4 @@
-SHELL=/bin/bash
+SHELL=/usr/bin/env bash
 # Makefile for OpenHands project
 
 # Variables

README.md

@@ -43,17 +43,17 @@ See the [Running OpenHands](https://docs.all-hands.dev/modules/usage/installatio
 system requirements and more information.
 
 ```bash
-docker pull docker.all-hands.dev/all-hands-ai/runtime:0.25-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.26-nikolaik
 
 docker run -it --rm --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-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:0.25
+    docker.all-hands.dev/all-hands-ai/openhands:0.26
 ```
 
 You'll find OpenHands running at [http://localhost:3000](http://localhost:3000)!

build.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -e
 
 poetry build -v

config.template.toml

@@ -83,9 +83,6 @@ workspace_base = "./workspace"
 # Runtime environment
 #runtime = "docker"
 
-# Runtime executor
-#runtime_executor = "openhands.runtime.executor:ActionExecutor"
-
 # Name of the default agent
 #default_agent = "CodeActAgent"
 

containers/build.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 # Initialize variables with default values

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:0.25-nikolaik}
+      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-ghcr.io/all-hands-ai/runtime:0.26-nikolaik}
       - SANDBOX_USER_ID=${SANDBOX_USER_ID:-1234}
       - WORKSPACE_MOUNT_PATH=${WORKSPACE_BASE:-$PWD/workspace}
     ports:

containers/dev/dev.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -o pipefail
 
 function get_docker() {

docker-compose.yml

@@ -7,7 +7,7 @@ services:
     image: openhands:latest
     container_name: openhands-app-${DATE:-}
     environment:
-      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-docker.all-hands.dev/all-hands-ai/runtime:0.25-nikolaik}
+      - SANDBOX_RUNTIME_CONTAINER_IMAGE=${SANDBOX_RUNTIME_CONTAINER_IMAGE:-docker.all-hands.dev/all-hands-ai/runtime:0.26-nikolaik}
       #- SANDBOX_USER_ID=${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:

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:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-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:0.25 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.26 \
     python -m openhands.core.cli
 ```
 

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:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-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:0.25 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.26 \
     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:0.25-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.26-nikolaik
 
 docker run -it --rm --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-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:0.25
+    docker.all-hands.dev/all-hands-ai/openhands:0.26
 ```
 
 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/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:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-nikolaik \
     -v /var/run/docker.sock:/var/run/docker.sock \
     # ...
 ```

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:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-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:0.25 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.26 \
     python -m openhands.core.cli
 ```
 

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:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-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:0.25 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.26 \
     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:0.25-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.26-nikolaik
 
 docker run -it --rm --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-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:0.25
+    docker.all-hands.dev/all-hands-ai/openhands:0.26
 ```
 
 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/runtimes.md

@@ -11,7 +11,7 @@
 
 ```
 docker run # ...
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-nikolaik \
     -v /var/run/docker.sock:/var/run/docker.sock \
     # ...
 ```

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:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-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:0.25 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.26 \
     python -m openhands.core.cli
 ```
 

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:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-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:0.25 \
+    docker.all-hands.dev/all-hands-ai/openhands:0.26 \
     python -m openhands.core.main -t "write a bash script that prints hi"
 ```
 

docs/modules/usage/installation.mdx

@@ -58,17 +58,17 @@ A system with a modern processor and a minimum of **4GB RAM** is recommended to
 The easiest way to run OpenHands is in Docker.
 
 ```bash
-docker pull docker.all-hands.dev/all-hands-ai/runtime:0.25-nikolaik
+docker pull docker.all-hands.dev/all-hands-ai/runtime:0.26-nikolaik
 
 docker run -it --rm --pull=always \
-    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-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:0.25
+    docker.all-hands.dev/all-hands-ai/openhands:0.26
 ```
 
 You'll find OpenHands running at http://localhost:3000!

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:0.25-nikolaik \
+    -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.26-nikolaik \
     -v /var/run/docker.sock:/var/run/docker.sock \
     # ...
 ```

evaluation/benchmarks/EDA/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/agent_bench/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/aider_bench/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/biocoder/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/bird/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/browsing_delegation/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/commit0_bench/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/discoverybench/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/gaia/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/gorilla/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/gpqa/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/humanevalfix/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/logic_reasoning/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/miniwob/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/mint/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/ml_bench/scripts/cleanup.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 # Step 1: Stop all running containers
 echo "Stopping all running containers..."

evaluation/benchmarks/ml_bench/scripts/run_analysis.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 RESULT_FILE=$1
 MODEL_CONFIG=$2

evaluation/benchmarks/ml_bench/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/scienceagentbench/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/swe_bench/scripts/docker/pull_all_eval_docker.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -e
 
 LEVEL=$1

evaluation/benchmarks/swe_bench/scripts/docker/push_eval_docker.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 # This is ONLY used for pushing docker images created by https://github.com/princeton-nlp/SWE-bench/blob/main/docs/20240627_docker/README.md
 

evaluation/benchmarks/swe_bench/scripts/eval/convert_oh_folder_to_swebench_submission.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 FOLDER_PATH=$1
 NEW_FOLDER_PATH=${FOLDER_PATH}.swebench_submission

evaluation/benchmarks/swe_bench/scripts/eval_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 PROCESS_FILEPATH=$1
 if [ -z "$PROCESS_FILEPATH" ]; then

evaluation/benchmarks/swe_bench/scripts/eval_infer_remote.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 INPUT_FILE=$1

evaluation/benchmarks/swe_bench/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/swe_bench/scripts/setup/instance_swe_entry.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 source ~/.bashrc
 SWEUTIL_DIR=/swe_util

evaluation/benchmarks/swe_bench/scripts/setup/prepare_swe_utils.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 set -e
 EVAL_WORKSPACE="evaluation/benchmarks/swe_bench/eval_workspace"

evaluation/benchmarks/swe_bench/scripts/setup/swe_entry.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 set -e
 

evaluation/benchmarks/the_agent_company/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 ##################################################################################################
 # Adapted from https://github.com/TheAgentCompany/TheAgentCompany/blob/main/evaluation/run_eval.sh

evaluation/benchmarks/toolqa/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/visualwebarena/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/benchmarks/webarena/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/integration_tests/scripts/run_infer.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 set -eo pipefail
 
 source "evaluation/utils/version_control.sh"

evaluation/regression/cases/hello-name/start/hello_world.sh

@@ -1,2 +1,2 @@
-#!/bin/bash
+#!/usr/bin/env bash
 echo "hello world"

evaluation/utils/scripts/cleanup_remote_runtime.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 
 # API base URL

frontend/__tests__/utils/extract-model-and-provider.test.ts

@@ -65,6 +65,12 @@ describe("extractModelAndProvider", () => {
       separator: "/",
     });
 
+    expect(extractModelAndProvider("claude-3-7-sonnet-20250219")).toEqual({
+      provider: "anthropic",
+      model: "claude-3-7-sonnet-20250219",
+      separator: "/",
+    });
+
     expect(extractModelAndProvider("claude-3-haiku-20240307")).toEqual({
       provider: "anthropic",
       model: "claude-3-haiku-20240307",

frontend/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "openhands-frontend",
-  "version": "0.25.0",
+  "version": "0.26.0",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "openhands-frontend",
-      "version": "0.25.0",
+      "version": "0.26.0",
       "dependencies": {
         "@heroui/react": "2.6.14",
         "@monaco-editor/react": "^4.7.0-rc.0",

frontend/package.json

@@ -1,6 +1,6 @@
 {
   "name": "openhands-frontend",
-  "version": "0.25.0",
+  "version": "0.26.0",
   "private": true,
   "type": "module",
   "engines": {

frontend/src/utils/verified-models.ts

@@ -3,6 +3,7 @@ export const VERIFIED_PROVIDERS = ["openai", "azure", "anthropic", "deepseek"];
 export const VERIFIED_MODELS = [
   "o3-mini-2025-01-31",
   "claude-3-5-sonnet-20241022",
+  "claude-3-7-sonnet-20250219",
   "deepseek-chat",
 ];
 
@@ -31,4 +32,5 @@ export const VERIFIED_ANTHROPIC_MODELS = [
   "claude-3-haiku-20240307",
   "claude-3-opus-20240229",
   "claude-3-sonnet-20240229",
+  "claude-3-7-sonnet-20250219",
 ];

openhands/core/config/__init__.py

@@ -5,6 +5,7 @@ from openhands.core.config.config_utils import (
     OH_MAX_ITERATIONS,
     get_field_info,
 )
+from openhands.core.config.extended_config import ExtendedConfig
 from openhands.core.config.llm_config import LLMConfig
 from openhands.core.config.sandbox_config import SandboxConfig
 from openhands.core.config.security_config import SecurityConfig
@@ -28,6 +29,7 @@ __all__ = [
     'LLMConfig',
     'SandboxConfig',
     'SecurityConfig',
+    'ExtendedConfig',
     'load_app_config',
     'load_from_env',
     'load_from_toml',

openhands/core/config/app_config.py

@@ -9,6 +9,7 @@ from openhands.core.config.config_utils import (
     OH_MAX_ITERATIONS,
     model_defaults_to_dict,
 )
+from openhands.core.config.extended_config import ExtendedConfig
 from openhands.core.config.llm_config import LLMConfig
 from openhands.core.config.sandbox_config import SandboxConfig
 from openhands.core.config.security_config import SecurityConfig
@@ -52,6 +53,7 @@ class AppConfig(BaseModel):
     default_agent: str = Field(default=OH_DEFAULT_AGENT)
     sandbox: SandboxConfig = Field(default_factory=SandboxConfig)
     security: SecurityConfig = Field(default_factory=SecurityConfig)
+    extended: ExtendedConfig = Field(default_factory=lambda: ExtendedConfig({}))
     runtime: str = Field(default='docker')
     file_store: str = Field(default='local')
     file_store_path: str = Field(default='/tmp/openhands_file_store')

openhands/core/config/extended_config.py

@@ -0,0 +1,40 @@
+from pydantic import RootModel
+
+
+class ExtendedConfig(RootModel[dict]):
+    """Configuration for extended functionalities.
+
+    This is implemented as a root model so that the entire input is stored
+    as the root value. This allows arbitrary keys to be stored and later
+    accessed via attribute or dictionary-style access.
+    """
+
+    @property
+    def root(self) -> dict:  # type annotation to help mypy
+        return super().root
+
+    def __str__(self) -> str:
+        # Use the root dict to build a string representation.
+        attr_str = [f'{k}={repr(v)}' for k, v in self.root.items()]
+        return f"ExtendedConfig({', '.join(attr_str)})"
+
+    def __repr__(self) -> str:
+        return self.__str__()
+
+    @classmethod
+    def from_dict(cls, data: dict) -> 'ExtendedConfig':
+        # Create an instance directly by wrapping the input dict.
+        return cls(data)
+
+    def __getitem__(self, key: str) -> object:
+        # Provide dictionary-like access via the root dict.
+        return self.root[key]
+
+    def __getattr__(self, key: str) -> object:
+        # Fallback for attribute access using the root dict.
+        try:
+            return self.root[key]
+        except KeyError as e:
+            raise AttributeError(
+                f"'ExtendedConfig' object has no attribute '{key}'"
+            ) from e

openhands/core/config/sandbox_config.py

@@ -53,11 +53,11 @@ class SandboxConfig(BaseModel):
     remote_runtime_api_timeout: int = Field(default=10)
     remote_runtime_enable_retries: bool = Field(default=False)
     remote_runtime_class: str | None = Field(
-        default='sysbox'
+        default=None
     )  # can be "None" (default to gvisor) or "sysbox" (support docker inside runtime + more stable)
     enable_auto_lint: bool = Field(
-        default=False  # once enabled, OpenHands would lint files after editing
-    )
+        default=False
+    )  # once enabled, OpenHands would lint files after editing
     use_host_network: bool = Field(default=False)
     runtime_extra_build_args: list[str] | None = Field(default=None)
     initialize_plugins: bool = Field(default=True)

openhands/core/config/utils.py

@@ -19,6 +19,7 @@ from openhands.core.config.config_utils import (
     OH_DEFAULT_AGENT,
     OH_MAX_ITERATIONS,
 )
+from openhands.core.config.extended_config import ExtendedConfig
 from openhands.core.config.llm_config import LLMConfig
 from openhands.core.config.sandbox_config import SandboxConfig
 from openhands.core.config.security_config import SecurityConfig
@@ -134,6 +135,10 @@ def load_from_toml(cfg: AppConfig, toml_file: str = 'config.toml') -> None:
     for key, value in toml_config.items():
         if isinstance(value, dict):
             try:
+                if key.lower() == 'extended':
+                    # For ExtendedConfig (RootModel), pass the entire dict as the root value
+                    cfg.extended = ExtendedConfig(value)
+                    continue
                 if key is not None and key.lower() == 'agent':
                     # Every entry here is either a field for the default `agent` config group, or itself a group
                     # The best way to tell the difference is to try to parse it as an AgentConfig object

openhands/llm/llm.py

@@ -8,7 +8,6 @@ from typing import Any, Callable
 import requests
 
 from openhands.core.config import LLMConfig
-from openhands.utils.ensure_httpx_close import EnsureHttpxClose
 
 with warnings.catch_warnings():
     warnings.simplefilter('ignore')
@@ -43,6 +42,7 @@ LLM_RETRY_EXCEPTIONS: tuple[type[Exception], ...] = (RateLimitError,)
 # cache prompt supporting models
 # remove this when we gemini and deepseek are supported
 CACHE_PROMPT_SUPPORTED_MODELS = [
+    'claude-3-7-sonnet-20250219',
     'claude-3-5-sonnet-20241022',
     'claude-3-5-sonnet-20240620',
     'claude-3-5-haiku-20241022',
@@ -52,6 +52,7 @@ CACHE_PROMPT_SUPPORTED_MODELS = [
 
 # function calling supporting models
 FUNCTION_CALLING_SUPPORTED_MODELS = [
+    'claude-3-7-sonnet-20250219',
     'claude-3-5-sonnet',
     'claude-3-5-sonnet-20240620',
     'claude-3-5-sonnet-20241022',
@@ -231,9 +232,9 @@ class LLM(RetryMixin, DebugMixin):
 
             # Record start time for latency measurement
             start_time = time.time()
-            with EnsureHttpxClose():
-                # we don't support streaming here, thus we get a ModelResponse
-                resp: ModelResponse = self._completion_unwrapped(*args, **kwargs)
+
+            # we don't support streaming here, thus we get a ModelResponse
+            resp: ModelResponse = self._completion_unwrapped(*args, **kwargs)
 
             # Calculate and record latency
             latency = time.time() - start_time
@@ -288,11 +289,7 @@ class LLM(RetryMixin, DebugMixin):
                     'messages': messages,
                     'response': resp,
                     'args': args,
-                    'kwargs': {
-                        k: v
-                        for k, v in kwargs.items()
-                        if k not in ('messages', 'client')
-                    },
+                    'kwargs': {k: v for k, v in kwargs.items() if k != 'messages'},
                     'timestamp': time.time(),
                     'cost': cost,
                 }

openhands/runtime/action_execution_server.py

@@ -15,7 +15,7 @@ import tempfile
 import time
 import traceback
 from contextlib import asynccontextmanager
-from typing import Type
+from pathlib import Path
 from zipfile import ZipFile
 
 from fastapi import Depends, FastAPI, HTTPException, Request, UploadFile
@@ -60,6 +60,7 @@ from openhands.runtime.utils.files import insert_lines, read_lines
 from openhands.runtime.utils.memory_monitor import MemoryMonitor
 from openhands.runtime.utils.runtime_init import init_user_and_working_directory
 from openhands.runtime.utils.system_stats import get_system_stats
+from openhands.utils.async_utils import call_sync_from_async, wait_all
 
 
 class ActionRequest(BaseModel):
@@ -67,6 +68,7 @@ class ActionRequest(BaseModel):
 
 
 ROOT_GID = 0
+
 SESSION_API_KEY = os.environ.get('SESSION_API_KEY')
 api_key_header = APIKeyHeader(name='X-Session-API-Key', auto_error=False)
 
@@ -133,6 +135,7 @@ class ActionExecutor:
     """ActionExecutor is running inside docker sandbox.
     It is responsible for executing actions received from OpenHands backend and producing observations.
     """
+
     def __init__(
         self,
         plugins_to_load: list[Plugin],
@@ -460,7 +463,6 @@ class ActionExecutor:
         if self.bash_session is not None:
             self.bash_session.close()
         self.browser.close()
->>>>>>> origin/main
 
 
 if __name__ == '__main__':
@@ -478,12 +480,6 @@ if __name__ == '__main__':
         help='BrowserGym environment used for browser evaluation',
         default=None,
     )
-    parser.add_argument(
-        '--executor-class',
-        type=str,
-        default='openhands.runtime.executor:ActionExecutor',
-        help='Action executor class to use (format: module.path:ClassName)',
-    )
     # example: python client.py 8000 --working-dir /workspace --plugins JupyterRequirement
     args = parser.parse_args()
 
@@ -494,13 +490,12 @@ if __name__ == '__main__':
                 raise ValueError(f'Plugin {plugin} not found')
             plugins_to_load.append(ALL_PLUGINS[plugin]())  # type: ignore
 
-    executor_class = get_action_executor_class(args.executor_class)
-    client: RuntimeExecutor | None = None
+    client: ActionExecutor | None = None
 
     @asynccontextmanager
     async def lifespan(app: FastAPI):
         global client
-        client = executor_class(
+        client = ActionExecutor(
             plugins_to_load,
             work_dir=args.working_dir,
             username=args.username,

openhands/runtime/executor/__init__.py

@@ -1,4 +0,0 @@
-from .base import RuntimeExecutor
-from .action_executor import ActionExecutor, BaseActionExecutor
-
-__all__ = ['ActionExecutor', 'BaseActionExecutor', 'RuntimeExecutor']

openhands/runtime/executor/action_executor.py

@@ -1,267 +0,0 @@
-import base64
-import json
-import mimetypes
-import os
-from pathlib import Path
-import re
-from openhands_aci.utils.diff import get_diff
-from openhands.core.logger import openhands_logger as logger
-from openhands.events.action.browse import BrowseInteractiveAction, BrowseURLAction
-from openhands.events.action.commands import IPythonRunCellAction
-from openhands.events.action.files import FileReadAction, FileWriteAction
-from openhands.events.event import FileEditSource, FileReadSource
-from openhands.events.observation.commands import (
-    IPythonRunCellObservation,
-)
-from openhands.events.observation.error import ErrorObservation
-from openhands.events.observation.files import (
-    FileEditObservation,
-    FileReadObservation,
-    FileWriteObservation,
-)
-from openhands.events.observation.observation import Observation
-from openhands.runtime.browser import browse
-from openhands.runtime.executor.base import RuntimeExecutor
-from openhands.runtime.plugins.jupyter import JupyterPlugin
-from openhands.runtime.utils.files import insert_lines, read_lines
-
-
-class BaseActionExecutor(RuntimeExecutor):
-    """Runtime executor that dynamically dispatches actions to the appropriate method based on their name."""
-
-    async def run_action(self, action) -> Observation:
-        async with self.lock:
-            action_type = action.action
-            logger.debug(f'Running action:\n{action}')
-            observation = await getattr(self, action_type)(action)
-            logger.debug(f'Action output:\n{observation}')
-            return observation
-
-
-class ActionExecutor(BaseActionExecutor):
-    """ActionExecutor runs inside docker sandbox.
-    It is responsible for executing actions received from OpenHands backend and producing observations.
-    It is a BaseActionExectuor that provides a default implementation for all of the built-in actions.
-    """
-
-    async def run_ipython(self, action: IPythonRunCellAction) -> Observation:
-        assert self.bash_session is not None
-        if 'jupyter' in self.plugins:
-            _jupyter_plugin: JupyterPlugin = self.plugins['jupyter']  # type: ignore
-            # This is used to make AgentSkills in Jupyter aware of the
-            # current working directory in Bash
-            jupyter_cwd = getattr(self, '_jupyter_cwd', None)
-            if self.bash_session.cwd != jupyter_cwd:
-                logger.debug(
-                    f'{self.bash_session.cwd} != {jupyter_cwd} -> reset Jupyter PWD'
-                )
-                reset_jupyter_cwd_code = (
-                    f'import os; os.chdir("{self.bash_session.cwd}")'
-                )
-                _aux_action = IPythonRunCellAction(code=reset_jupyter_cwd_code)
-                _reset_obs: IPythonRunCellObservation = await _jupyter_plugin.run(
-                    _aux_action
-                )
-                logger.debug(
-                    f'Changed working directory in IPython to: {self.bash_session.cwd}. Output: {_reset_obs}'
-                )
-                self._jupyter_cwd = self.bash_session.cwd
-
-            obs: IPythonRunCellObservation = await _jupyter_plugin.run(action)
-            obs.content = obs.content.rstrip()
-            matches = re.findall(
-                r'<oh_aci_output_[0-9a-f]{32}>(.*?)</oh_aci_output_[0-9a-f]{32}>',
-                obs.content,
-                re.DOTALL,
-            )
-            if matches:
-                results: list[str] = []
-                if len(matches) == 1:
-                    # Use specific actions/observations types
-                    match = matches[0]
-                    try:
-                        result_dict = json.loads(match)
-                        if result_dict.get('path'):  # Successful output
-                            if (
-                                result_dict['new_content'] is not None
-                            ):  # File edit commands
-                                diff = get_diff(
-                                    old_contents=result_dict['old_content']
-                                    or '',  # old_content is None when file is created
-                                    new_contents=result_dict['new_content'],
-                                    filepath=result_dict['path'],
-                                )
-                                return FileEditObservation(
-                                    content=diff,
-                                    path=result_dict['path'],
-                                    old_content=result_dict['old_content'],
-                                    new_content=result_dict['new_content'],
-                                    prev_exist=result_dict['prev_exist'],
-                                    impl_source=FileEditSource.OH_ACI,
-                                    formatted_output_and_error=result_dict[
-                                        'formatted_output_and_error'
-                                    ],
-                                )
-                            else:  # File view commands
-                                return FileReadObservation(
-                                    content=result_dict['formatted_output_and_error'],
-                                    path=result_dict['path'],
-                                    impl_source=FileReadSource.OH_ACI,
-                                )
-                        else:  # Error output
-                            results.append(result_dict['formatted_output_and_error'])
-                    except json.JSONDecodeError:
-                        # Handle JSON decoding errors if necessary
-                        results.append(
-                            f"Invalid JSON in 'openhands-aci' output: {match}"
-                        )
-                else:
-                    for match in matches:
-                        try:
-                            result_dict = json.loads(match)
-                            results.append(result_dict['formatted_output_and_error'])
-                        except json.JSONDecodeError:
-                            # Handle JSON decoding errors if necessary
-                            results.append(
-                                f"Invalid JSON in 'openhands-aci' output: {match}"
-                            )
-
-                # Combine the results (e.g., join them) or handle them as required
-                obs.content = '\n'.join(str(result) for result in results)
-
-            if action.include_extra:
-                obs.content += (
-                    f'\n[Jupyter current working directory: {self.bash_session.cwd}]'
-                )
-                obs.content += f'\n[Jupyter Python interpreter: {_jupyter_plugin.python_interpreter_path}]'
-            return obs
-        else:
-            raise RuntimeError(
-                'JupyterRequirement not found. Unable to run IPython action.'
-            )
-
-    def _resolve_path(self, path: str, working_dir: str) -> str:
-        filepath = Path(path)
-        if not filepath.is_absolute():
-            return str(Path(working_dir) / filepath)
-        return str(filepath)
-
-    async def read(self, action: FileReadAction) -> Observation:
-        assert self.bash_session is not None
-        if action.impl_source == FileReadSource.OH_ACI:
-            return await self.run_ipython(
-                IPythonRunCellAction(
-                    code=action.translated_ipython_code,
-                    include_extra=False,
-                )
-            )
-
-        # NOTE: the client code is running inside the sandbox,
-        # so there's no need to check permission
-        working_dir = self.bash_session.cwd
-        filepath = self._resolve_path(action.path, working_dir)
-        try:
-            if filepath.lower().endswith(('.png', '.jpg', '.jpeg', '.bmp', '.gif')):
-                with open(filepath, 'rb') as file:
-                    image_data = file.read()
-                    encoded_image = base64.b64encode(image_data).decode('utf-8')
-                    mime_type, _ = mimetypes.guess_type(filepath)
-                    if mime_type is None:
-                        mime_type = 'image/png'  # default to PNG if mime type cannot be determined
-                    encoded_image = f'data:{mime_type};base64,{encoded_image}'
-
-                return FileReadObservation(path=filepath, content=encoded_image)
-            elif filepath.lower().endswith('.pdf'):
-                with open(filepath, 'rb') as file:
-                    pdf_data = file.read()
-                    encoded_pdf = base64.b64encode(pdf_data).decode('utf-8')
-                    encoded_pdf = f'data:application/pdf;base64,{encoded_pdf}'
-                return FileReadObservation(path=filepath, content=encoded_pdf)
-            elif filepath.lower().endswith(('.mp4', '.webm', '.ogg')):
-                with open(filepath, 'rb') as file:
-                    video_data = file.read()
-                    encoded_video = base64.b64encode(video_data).decode('utf-8')
-                    mime_type, _ = mimetypes.guess_type(filepath)
-                    if mime_type is None:
-                        mime_type = 'video/mp4'  # default to MP4 if MIME type cannot be determined
-                    encoded_video = f'data:{mime_type};base64,{encoded_video}'
-
-                return FileReadObservation(path=filepath, content=encoded_video)
-
-            with open(filepath, 'r', encoding='utf-8') as file:
-                lines = read_lines(file.readlines(), action.start, action.end)
-        except FileNotFoundError:
-            return ErrorObservation(
-                f'File not found: {filepath}. Your current working directory is {working_dir}.'
-            )
-        except UnicodeDecodeError:
-            return ErrorObservation(f'File could not be decoded as utf-8: {filepath}.')
-        except IsADirectoryError:
-            return ErrorObservation(
-                f'Path is a directory: {filepath}. You can only read files'
-            )
-
-        code_view = ''.join(lines)
-        return FileReadObservation(path=filepath, content=code_view)
-
-    async def write(self, action: FileWriteAction) -> Observation:
-        assert self.bash_session is not None
-        working_dir = self.bash_session.cwd
-        filepath = self._resolve_path(action.path, working_dir)
-
-        insert = action.content.split('\n')
-        try:
-            if not os.path.exists(os.path.dirname(filepath)):
-                os.makedirs(os.path.dirname(filepath))
-
-            file_exists = os.path.exists(filepath)
-            if file_exists:
-                file_stat = os.stat(filepath)
-            else:
-                file_stat = None
-
-            mode = 'w' if not file_exists else 'r+'
-            try:
-                with open(filepath, mode, encoding='utf-8') as file:
-                    if mode != 'w':
-                        all_lines = file.readlines()
-                        new_file = insert_lines(
-                            insert, all_lines, action.start, action.end
-                        )
-                    else:
-                        new_file = [i + '\n' for i in insert]
-
-                    file.seek(0)
-                    file.writelines(new_file)
-                    file.truncate()
-
-                # Handle file permissions
-                if file_exists:
-                    assert file_stat is not None
-                    # restore the original file permissions if the file already exists
-                    os.chmod(filepath, file_stat.st_mode)
-                    os.chown(filepath, file_stat.st_uid, file_stat.st_gid)
-                else:
-                    # set the new file permissions if the file is new
-                    os.chmod(filepath, 0o664)
-                    os.chown(filepath, self.user_id, self.user_id)
-
-            except FileNotFoundError:
-                return ErrorObservation(f'File not found: {filepath}')
-            except IsADirectoryError:
-                return ErrorObservation(
-                    f'Path is a directory: {filepath}. You can only write to files'
-                )
-            except UnicodeDecodeError:
-                return ErrorObservation(
-                    f'File could not be decoded as utf-8: {filepath}'
-                )
-        except PermissionError:
-            return ErrorObservation(f'Malformed paths not permitted: {filepath}')
-        return FileWriteObservation(content='', path=filepath)
-
-    async def browse(self, action: BrowseURLAction) -> Observation:
-        return await browse(action, self.browser)
-
-    async def browse_interactive(self, action: BrowseInteractiveAction) -> Observation:
-        return await browse(action, self.browser)

openhands/runtime/executor/base.py

@@ -1,126 +0,0 @@
-import asyncio
-import time
-from openhands.core.logger import openhands_logger as logger
-from openhands.events.action.commands import CmdRunAction, IPythonRunCellAction
-from openhands.events.observation.commands import CmdOutputObservation
-from openhands.events.observation.error import ErrorObservation
-from openhands.runtime.browser.browser_env import BrowserEnv
-from openhands.runtime.plugins.jupyter import JupyterPlugin
-from openhands.runtime.plugins.requirement import Plugin
-from openhands.runtime.utils.bash import BashSession
-from openhands.runtime.utils.runtime_init import init_user_and_working_directory
-from openhands.utils.async_utils import call_sync_from_async, wait_all
-
-
-ROOT_GID = 0
-INIT_COMMANDS = [
-    'git config --global user.name "openhands" && git config --global user.email "openhands@all-hands.dev" && alias git="git --no-pager"',
-]
-
-
-class RuntimeExecutor:
-    """RuntimeExecutor for running inside docker sandbox.
-    It provides a minimal base class that handles initialization of the executor, and provides a run method to execute bash commands.
-    """
-
-    def __init__(
-        self,
-        plugins_to_load: list[Plugin],
-        work_dir: str,
-        username: str,
-        user_id: int,
-        browsergym_eval_env: str | None,
-    ) -> None:
-        self.plugins_to_load = plugins_to_load
-        self._initial_cwd = work_dir
-        self.username = username
-        self.user_id = user_id
-        _updated_user_id = init_user_and_working_directory(
-            username=username, user_id=self.user_id, initial_cwd=work_dir
-        )
-        if _updated_user_id is not None:
-            self.user_id = _updated_user_id
-
-        self.bash_session: BashSession | None = None
-        self.lock = asyncio.Lock()
-        self.plugins: dict[str, Plugin] = {}
-        self.browser = BrowserEnv(browsergym_eval_env)
-        self.start_time = time.time()
-        self.last_execution_time = self.start_time
-        self._initialized = False
-
-    @property
-    def initial_cwd(self):
-        return self._initial_cwd
-
-    async def ainit(self):
-        # bash needs to be initialized first
-        self.bash_session = BashSession(
-            work_dir=self._initial_cwd,
-            username=self.username,
-        )
-        self.bash_session.initialize()
-        await wait_all(
-            (self._init_plugin(plugin) for plugin in self.plugins_to_load),
-            timeout=30,
-        )
-
-        # This is a temporary workaround
-        # TODO: refactor AgentSkills to be part of JupyterPlugin
-        # AFTER ServerRuntime is deprecated
-        if 'agent_skills' in self.plugins and 'jupyter' in self.plugins:
-            obs = await self.run_ipython(
-                IPythonRunCellAction(
-                    code='from openhands.runtime.plugins.agent_skills.agentskills import *\n'
-                )
-            )
-            logger.debug(f'AgentSkills initialized: {obs}')
-
-        await self._init_bash_commands()
-        logger.debug('Runtime client initialized.')
-
-        self._initialized = True
-
-    @property
-    def initialized(self) -> bool:
-        return self._initialized
-
-    async def _init_plugin(self, plugin: Plugin):
-        assert self.bash_session is not None
-        await plugin.initialize(self.username)
-        self.plugins[plugin.name] = plugin
-        logger.debug(f'Initializing plugin: {plugin.name}')
-
-        if isinstance(plugin, JupyterPlugin):
-            await self.run_ipython(
-                IPythonRunCellAction(
-                    code=f'import os; os.chdir("{self.bash_session.cwd}")'
-                )
-            )
-
-    async def _init_bash_commands(self):
-        logger.debug(f'Initializing by running {len(INIT_COMMANDS)} bash commands...')
-        for command in INIT_COMMANDS:
-            action = CmdRunAction(command=command)
-            action.timeout = 300
-            logger.debug(f'Executing init command: {command}')
-            obs = await self.run(action)
-            assert isinstance(obs, CmdOutputObservation)
-            logger.debug(
-                f'Init command outputs (exit code: {obs.exit_code}): {obs.content}'
-            )
-            assert obs.exit_code == 0
-
-        logger.debug('Bash init commands completed')
-
-    async def run(
-        self, action: CmdRunAction
-    ) -> CmdOutputObservation | ErrorObservation:
-        assert self.bash_session is not None
-        obs = await call_sync_from_async(self.bash_session.execute, action)
-        return obs
-
-    def close(self):
-        if self.bash_session is not None:
-            self.bash_session.close()
-        self.browser.close()

openhands/runtime/utils/command.py

@@ -55,8 +55,6 @@ def get_action_execution_server_startup_command(
         '--user-id',
         str(user_id),
         *browsergym_args,
-        '--executor-class',
-        app_config.runtime_executor,
     ]
 
     return base_cmd

openhands/utils/ensure_httpx_close.py

@@ -1,43 +0,0 @@
-"""
-LiteLLM currently have an issue where HttpHandlers are being created but not
-closed. We have submitted a PR to them, (https://github.com/BerriAI/litellm/pull/8711)
-and their dev team say they are in the process of a refactor that will fix this, but
-in the meantime, we need to manage the lifecycle of the httpx.Client manually.
-
-We can't simply pass in our own client object, because all the different implementations use
-different types of client object.
-
-So we monkey patch the httpx.Client class to track newly created instances and close these
-when the operations complete. (This is relatively safe, as if the client is reused after this
-then is will transparently reopen)
-
-Hopefully, this will be fixed soon and we can remove this abomination.
-"""
-
-from dataclasses import dataclass, field
-from functools import wraps
-from typing import Callable
-
-from httpx import Client
-
-
-@dataclass
-class EnsureHttpxClose:
-    clients: list[Client] = field(default_factory=list)
-    original_init: Callable | None = None
-
-    def __enter__(self):
-        self.original_init = Client.__init__
-
-        @wraps(Client.__init__)
-        def init_wrapper(*args, **kwargs):
-            self.clients.append(args[0])
-            return self.original_init(*args, **kwargs)  # type: ignore
-
-        Client.__init__ = init_wrapper
-
-    def __exit__(self, type, value, traceback):
-        Client.__init__ = self.original_init
-        while self.clients:
-            client = self.clients.pop()
-            client.close()

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "openhands-ai"
-version = "0.25.0"
+version = "0.26.0"
 description = "OpenHands: Code Less, Make More"
 authors = ["OpenHands"]
 license = "MIT"
@@ -109,7 +109,6 @@ reportlab = "*"
 [tool.coverage.run]
 concurrency = ["gevent"]
 
-
 [tool.poetry.group.runtime.dependencies]
 jupyterlab = "*"
 notebook = "*"
@@ -138,7 +137,6 @@ ignore = ["D1"]
 [tool.ruff.lint.pydocstyle]
 convention = "google"
 
-
 [tool.poetry.group.evaluation.dependencies]
 streamlit = "*"
 whatthepatch = "*"

tests/unit/test_config_extended.py

@@ -0,0 +1,169 @@
+import os
+
+import pytest
+
+from openhands.core.config.app_config import AppConfig
+from openhands.core.config.extended_config import ExtendedConfig
+from openhands.core.config.utils import load_from_toml
+
+
+def test_extended_config_from_dict():
+    """
+    Test that ExtendedConfig.from_dict successfully creates an instance
+    from a dictionary containing arbitrary extra keys.
+    """
+    data = {'foo': 'bar', 'baz': 123, 'flag': True}
+    ext_cfg = ExtendedConfig.from_dict(data)
+
+    # Check that the keys are accessible both as attributes and via __getitem__
+    assert ext_cfg.foo == 'bar'
+    assert ext_cfg['baz'] == 123
+    assert ext_cfg.flag is True
+    # Verify the root dictionary contains all keys
+    assert ext_cfg.root == data
+
+
+def test_extended_config_empty():
+    """
+    Test that an empty ExtendedConfig can be created and accessed.
+    """
+    ext_cfg = ExtendedConfig.from_dict({})
+    assert ext_cfg.root == {}
+
+    # Creating directly should also work
+    ext_cfg2 = ExtendedConfig({})
+    assert ext_cfg2.root == {}
+
+
+def test_extended_config_str_and_repr():
+    """
+    Test that __str__ and __repr__ return the correct string representations
+    of the ExtendedConfig instance.
+    """
+    data = {'alpha': 'test', 'beta': 42}
+    ext_cfg = ExtendedConfig.from_dict(data)
+    string_repr = str(ext_cfg)
+    repr_str = repr(ext_cfg)
+
+    # Ensure the representations include our key/value pairs
+    assert "alpha='test'" in string_repr
+    assert 'beta=42' in string_repr
+
+    # __repr__ should match __str__
+    assert string_repr == repr_str
+
+
+def test_extended_config_getitem_and_getattr():
+    """
+    Test that __getitem__ and __getattr__ can be used to access values
+    in the ExtendedConfig instance.
+    """
+    data = {'key1': 'value1', 'key2': 2}
+    ext_cfg = ExtendedConfig.from_dict(data)
+
+    # Attribute access
+    assert ext_cfg.key1 == 'value1'
+    # Dictionary-style access
+    assert ext_cfg['key2'] == 2
+
+
+def test_extended_config_invalid_key():
+    """
+    Test that accessing a non-existent key via attribute access raises AttributeError.
+    """
+    data = {'existing': 'yes'}
+    ext_cfg = ExtendedConfig.from_dict(data)
+
+    with pytest.raises(AttributeError):
+        _ = ext_cfg.nonexistent
+
+    with pytest.raises(KeyError):
+        _ = ext_cfg['nonexistent']
+
+
+def test_app_config_extended_from_toml(tmp_path: os.PathLike) -> None:
+    """
+    Test that the [extended] section in a TOML file is correctly loaded into
+    AppConfig.extended and that it accepts arbitrary keys.
+    """
+    # Create a temporary TOML file with multiple sections including [extended]
+    config_content = """
+[core]
+workspace_base = "/tmp/workspace"
+
+[llm]
+model = "test-model"
+api_key = "toml-api-key"
+
+[extended]
+custom1 = "custom_value"
+custom2 = 42
+llm = "overridden"  # even a key like 'llm' is accepted in extended
+
+[agent]
+memory_enabled = true
+"""
+    config_file = tmp_path / 'config.toml'
+    config_file.write_text(config_content)
+
+    # Load the TOML into the AppConfig instance
+    config = AppConfig()
+    load_from_toml(config, str(config_file))
+
+    # Verify that extended section is applied
+    assert config.extended.custom1 == 'custom_value'
+    assert config.extended.custom2 == 42
+    # Even though 'llm' is defined in extended, it should not affect the main llm config.
+    assert config.get_llm_config().model == 'test-model'
+
+
+def test_app_config_extended_default(tmp_path: os.PathLike) -> None:
+    """
+    Test that if there is no [extended] section in the TOML file,
+    AppConfig.extended remains its default (empty) ExtendedConfig.
+    """
+    config_content = """
+[core]
+workspace_base = "/tmp/workspace"
+
+[llm]
+model = "test-model"
+api_key = "toml-api-key"
+
+[agent]
+memory_enabled = true
+"""
+    config_file = tmp_path / 'config.toml'
+    config_file.write_text(config_content)
+
+    config = AppConfig()
+    load_from_toml(config, str(config_file))
+
+    # Extended config should be empty
+    assert config.extended.root == {}
+
+
+def test_app_config_extended_random_keys(tmp_path: os.PathLike) -> None:
+    """
+    Test that the extended section accepts arbitrary keys,
+    including ones not defined in any schema.
+    """
+    config_content = """
+[core]
+workspace_base = "/tmp/workspace"
+
+[extended]
+random_key = "random_value"
+another_key = 3.14
+"""
+    config_file = tmp_path / 'config.toml'
+    config_file.write_text(config_content)
+
+    config = AppConfig()
+    load_from_toml(config, str(config_file))
+
+    # Verify that extended config holds the arbitrary keys with correct values.
+    assert config.extended.random_key == 'random_value'
+    assert config.extended.another_key == 3.14
+    # Verify the root dictionary contains all keys
+    assert config.extended.root == {'random_key': 'random_value', 'another_key': 3.14}

tests/unit/test_ensure_httpx_close.py

@@ -1,84 +0,0 @@
-from httpx import Client
-
-from openhands.utils.ensure_httpx_close import EnsureHttpxClose
-
-
-def test_ensure_httpx_close_basic():
-    """Test basic functionality of EnsureHttpxClose."""
-    clients = []
-    ctx = EnsureHttpxClose()
-    with ctx:
-        # Create a client - should be tracked
-        client = Client()
-        assert client in ctx.clients
-        assert len(ctx.clients) == 1
-        clients.append(client)
-
-    # After context exit, client should be closed
-    assert client.is_closed
-
-
-def test_ensure_httpx_close_multiple_clients():
-    """Test EnsureHttpxClose with multiple clients."""
-    ctx = EnsureHttpxClose()
-    with ctx:
-        client1 = Client()
-        client2 = Client()
-        assert len(ctx.clients) == 2
-        assert client1 in ctx.clients
-        assert client2 in ctx.clients
-
-    assert client1.is_closed
-    assert client2.is_closed
-
-
-def test_ensure_httpx_close_nested():
-    """Test nested usage of EnsureHttpxClose."""
-    outer_ctx = EnsureHttpxClose()
-    with outer_ctx:
-        client1 = Client()
-        assert client1 in outer_ctx.clients
-
-        inner_ctx = EnsureHttpxClose()
-        with inner_ctx:
-            client2 = Client()
-            assert client2 in inner_ctx.clients
-            # Since both contexts are using the same monkey-patched __init__,
-            # both contexts will track all clients created while they are active
-            assert client2 in outer_ctx.clients
-
-        # After inner context, client2 should be closed
-        assert client2.is_closed
-        # client1 should still be open since outer context is still active
-        assert not client1.is_closed
-
-    # After outer context, both clients should be closed
-    assert client1.is_closed
-    assert client2.is_closed
-
-
-def test_ensure_httpx_close_exception():
-    """Test EnsureHttpxClose when an exception occurs."""
-    client = None
-    ctx = EnsureHttpxClose()
-    try:
-        with ctx:
-            client = Client()
-            raise ValueError('Test exception')
-    except ValueError:
-        pass
-
-    # Client should be closed even if an exception occurred
-    assert client is not None
-    assert client.is_closed
-
-
-def test_ensure_httpx_close_restore_init():
-    """Test that the original __init__ is restored after context exit."""
-    original_init = Client.__init__
-    ctx = EnsureHttpxClose()
-    with ctx:
-        assert Client.__init__ != original_init
-
-    # Original __init__ should be restored
-    assert Client.__init__ == original_init

tests/unit/test_llm.py

@@ -1,6 +1,4 @@
 import copy
-import tempfile
-from pathlib import Path
 from unittest.mock import MagicMock, patch
 
 import pytest
@@ -491,27 +489,3 @@ def test_llm_token_usage(mock_litellm_completion, default_config):
     assert usage_entry_2['cache_read_tokens'] == 1
     assert usage_entry_2['cache_write_tokens'] == 3
     assert usage_entry_2['response_id'] == 'test-response-usage-2'
-
-
-@patch('openhands.llm.llm.litellm_completion')
-def test_completion_with_log_completions(mock_litellm_completion, default_config):
-    with tempfile.TemporaryDirectory() as temp_dir:
-        default_config.log_completions = True
-        default_config.log_completions_folder = temp_dir
-        mock_response = {
-            'choices': [{'message': {'content': 'This is a mocked response.'}}]
-        }
-        mock_litellm_completion.return_value = mock_response
-
-        test_llm = LLM(config=default_config)
-        response = test_llm.completion(
-            messages=[{'role': 'user', 'content': 'Hello!'}],
-            stream=False,
-            drop_params=True,
-        )
-        assert (
-            response['choices'][0]['message']['content'] == 'This is a mocked response.'
-        )
-        files = list(Path(temp_dir).iterdir())
-        # Expect a log to be generated
-        assert len(files) == 1