Build agents your way.
Extend OpenHands with code, not constraints.
An open SDK for building custom agents and workflows.
The OpenHands SDK lets teams create specialized agents, integrate internal tools, and orchestrate complex workflows programmatically.
Solve real edge cases.
Closed platforms force you into predefined behaviors. OpenHands gives you primitives, not prescriptions, so you can adapt agents to your codebase, tooling, and governance needs.
Key capabilities
- Programmatic task creation
- Custom agent behaviors
- Tool and API integrations
- Model selection and routing
- Fine-grained execution control
How teams use it
- Build domain-specific agents
- Integrate agents into your apps
- Create internal AI platforms
- Extend OpenHands beyond defaults
| Feature | OpenHands | Devin | Claude Code | Factory |
|---|---|---|---|---|
| Open-source |  |  | | |
| Model-agnostic | | | | |
| Sandbox runtime | | | | |
| Self-hosted | | | | |
| Coding Agent SDK | | | | |
FAQ's
The OpenHands SDK is a developer toolkit for building, customizing, and orchestrating AI coding agents. It allows teams to integrate OpenHands capabilities directly into their own systems and workflows.
The Large Codebase SDK is a specialized component that enables OpenHands to understand and operate on complex codebases. It maps dependencies, sequences changes correctly, and allows multiple agents to work in parallel without conflicts.
Developers can use the SDK to build custom agent workflows, automate engineering processes, and extend OpenHands to fit their specific infrastructure and development needs.
Yes. The SDK is designed to be model-agnostic, allowing teams to use different large language models depending on performance, cost, or compliance requirements.
The SDK is best suited for platform teams, infrastructure engineers, and organizations looking to deeply integrate AI-driven development into their systems and scale it across teams.
Start building with the SDK.
%201.svg)