OpenHands/software-agent-sdk

What it solves

OpenHands Software Agent SDK provides a standardized way to build agents that can interact with codebases. It simplifies the process of creating AI agents capable of performing software engineering tasks, ranging from simple one-off tasks like writing documentation to complex, multi-agent workflows for refactoring and rewriting code.

How it works

The SDK provides Python and REST APIs that allow developers to define agents, assign them tools (such as terminal access, file editors, and task trackers), and manage conversations. Agents can operate within a local workspace or run inside ephemeral, isolated environments like Docker or Kubernetes via an Agent Server.

Who it’s for

Developers and engineers who want to build custom AI software agents for routine maintenance, automated documentation, or large-scale code modifications.

Highlights

• Flexible Workspaces: Supports both local machine workspaces and remote, ephemeral environments (Docker/Kubernetes).
• Composable Architecture: Provides a set of core components (Agents, Tools, Conversations, Workspaces) that can be integrated into larger systems.
• Built-in Tools: Includes ready-to-use tools like TerminalTool, FileEditorTool, and TaskTrackerTool.
• Skill Marketplace: Supports public skills (e.g., for uv and deno) that agents can automatically detect and use based on repository markers.
• Production Ready: Serves as the engine for the OpenHands CLI and OpenHands Cloud.