OpenHands Agent Canvas
A self-hosted control center for coding agents: OpenHands, Claude Code, Codex and Gemini on local and remote backends
Install
npm install -g @openhands/agent-canvas
agent-canvasNo sandbox, the agent gets full filesystem access
This is third-party code. Review the repository files before installing.
What it does
Agent Canvas starts agent conversations and automations from one interface. It runs the open-source OpenHands agent out of the box and can also use Claude Code, Codex, Gemini or any Agent-Client Protocol agent. Agents run locally, in Docker, on VMs or in OpenHands Cloud, and automations run on a schedule or on webhooks, integrating with Slack, GitHub, Linear and Notion. Any LLM can be used.
Who it is for. Teams and developers who want always-on agents running on their own infrastructure.
Good fit when
- You want agents on a server that keep working when your laptop is closed
- You need recurring automations: Slack reports or breaking GitHub issues into tasks
- You use several agents and want to manage them from one place
- You need the agent isolated in a Docker sandbox
Not a fit when
- You need a simple terminal agent without a web UI and backend
- You cannot isolate the agent and must not grant full filesystem access
Example request
Set up an automation that collects open PRs every morning and posts a summary to our Slack channelLimitations
Without the Docker sandbox the agent has full access to the machine's filesystem. Requires Node.js 22.12 and uv, or Docker. The system spans several repositories: the SDK, the TypeScript client and the automation service live separately.
How to disable. Stop the agent-canvas process or Docker container and remove the package with npm uninstall -g @openhands/agent-canvas.
Security check
- Agents run arbitrary commands, with full filesystem access when not sandboxed
- Automations receive tokens for Slack, GitHub and other services
- Sends code to the chosen LLM provider
README in short
The repository README now describes OpenHands Agent Canvas, a self-hosted control center for agents and automations. It gives three ways to run it: via npm without a sandbox, in Docker and from source. The architecture is built on the OpenHands Agent Server REST API plus an automation server. It also explains which related repository owns which part.
FAQ
Can I use Claude Code instead of the OpenHands agent?
Yes, the README lists support for Claude Code, Codex, Gemini and any Agent-Client Protocol agent.
Can the frontend and backend run separately?
Yes, use agent-canvas --frontend-only and --backend-only.
Related
A self-improving agent from Nous Research with a TUI, messaging gateway, cron jobs and skills it writes itself
Anthropic's coding agent for the terminal, IDE and GitHub: understands your codebase, runs tasks and handles git
OpenAI's coding agent that runs locally in your terminal, with IDE and desktop versions
Google's open-source terminal agent on Gemini models: code work, shell, web search and MCP