OpenHands

OpenHands Agent Canvas is a self-hosted developer control center for coding agents and automations. You use it to start agent conversations and to define automations such as generating reports that post to Slack or decomposing GitHub issues into tasks. It runs the open-source OpenHands agent by default and can also drive Claude Code, Codex, Gemini, or any agent that speaks the Agent-Client Protocol (ACP).

The frontend connects to one or more Agent Servers, a REST API for running multiple agents on a single host. A server can run on your laptop without a sandbox, inside a Docker container with a mounted projects directory, on a dedicated machine or cloud VM, or on OpenHands Cloud and Enterprise infrastructure, and you switch between backends from the same UI. An Agent Server is often paired with an Automation Server that runs agents on a schedule or in response to webhook events.

The system spans several repositories: this one holds the Agent Canvas frontend and local-stack orchestration, software-agent-sdk holds the Python SDK, Agent Server, agents, and tools, typescript-client is the browser client for the server API, and automation holds automation definitions, scheduling, and run history.

Features

• Multiple backends: run agents locally, in Docker, on VMs, or in OpenHands Cloud, and switch between them without leaving the UI
• Bring your own agent: OpenHands, Claude Code, Codex, Gemini CLI, or any ACP-compatible agent
• Bring your own model: LLM profiles let you use any model provider
• Automations: scheduled or webhook-triggered workflows that integrate with Slack, GitHub, Linear, Notion, and Jira
• Prebuilt automations: Slack channel monitoring, Linear ticket handling, Jira incident triage, GitHub code review, and retrospective drafting
• Split launch modes: agent-canvas --frontend-only or --backend-only to run parts of the stack separately
• Team sharing: one Agent Server can serve a team for code review and dependency updates while personal agents run locally