OpenHands Agent
AI agent · Coding agent · Engineering & development by OpenHands (All Hands AI)
The open-source software development agent that autonomously solves tasks end-to-end.
OpenHands Agent (formerly OpenDevin) is an open-source autonomous software development agent designed for individual developers and engineering teams. It solves software development tasks end-to-end, including bug fixes, feature implementation, and repository maintenance. The agent works by combining a large language model with a specialized prompt and an action space that includes tools for writing code, executing bash commands, and browsing the web. It is model-agnostic, allowing users to power it with models like Claude, GPT-4o, or open weights like Qwen and DeepSeek. By running in containe
- Price
- From 0 USD
- Languages
- English
- License
- MIT License
- Autonomy
- End-to-end autonomous task execution
- Model Support
- Model-agnostic (Claude, OpenAI, Qwen, Llama, etc.)
- Runtime Environment
- Containerized sandbox
- Primary Language
- Python (SDK)
Overview
OpenHands Agent (formerly OpenDevin) is an open-source autonomous software development agent designed for individual developers and engineering teams. It solves software development tasks end-to-end, including bug fixes, feature implementation, and repository maintenance. The agent works by combining a large language model with a specialized prompt and an action space that includes tools for writing code, executing bash commands, and browsing the web. It is model-agnostic, allowing users to power it with models like Claude, GPT-4o, or open weights like Qwen and DeepSeek. By running in containerized sandboxes, it ensures that autonomous actions are isolated and secure while keeping humans in control of final reviews and merges.
Coding Capabilities
- End-to-end Task Execution: The agent can autonomously solve software development tasks end-to-end, from fixing bugs to building new features.
- Repository Management: It can initialize new repositories, create.gitignore files, and commit changes to version control automatically.
- Automated Testing and Verification: The agent can write and run tests using frameworks like vitest to verify its own code changes.
- Bug Debugging and Fixing: It can identify API limitations like pagination and modify code to handle them correctly.
- Feature Implementation: The agent can implement new features such as dark mode toggles by modifying both logic and CSS files.
Tools and Integrations
- Bash and Code Execution: The agent can execute Bash commands and run code within its environment to test implementations.
- Web Browsing: It can browse the web to gather information or documentation needed for a task.
- File System Access: The agent can read and write files directly within the workspace to perform refactors or updates.
- GitHub Integration: It integrates with GitHub to view open pull requests and manage repository activities.
Developer Control
- Human-in-the-loop Control: Engineers remain in control of review, approval, and merge while the agent handles the repetitive work.
- Transparent Documentation: The agent provides a detailed README and instructions for developers to run and verify the created code.
- Interactive Debugging: Users can provide feedback or error logs in a chat interface to put the agent back on the right track.,label
Pricing
- Open Source
- Free
Specifications
- License
- MIT License
- Autonomy
- End-to-end autonomous task execution
- Model Support
- Model-agnostic (Claude, OpenAI, Qwen, Llama, etc.)
- Runtime Environment
- Containerized sandbox
- Primary Language
- Python (SDK)
Added by AntSpace from public sources (openhands.dev).