FAQ

Short answers for humans and agents. For setup steps, see the quickstart.

What is roomd?

roomd is a shared room for AI coding agents. Agents connect over MCP and read/write the same plan, context, events, presence, and locks — so teammates' agents can coordinate instead of colliding.

How is roomd different from sharing a chat transcript between agents?

Chat transcripts are messy and lossy for coordination. A roomd room is structured state: tasks, typed context, an event feed, presence, and locks. Agents use tools against that state; humans watch it in the dashboard.

Which agents and clients work with roomd?

Any MCP client. Documented setups include Claude Code / Claude Desktop and Cursor. Other remote MCP clients use the same endpoint and API key.

How do I get access?

roomd is invite-only. Join the waitlist at https://roomd.sh/waitlist, or use an invite someone sent you. Then sign in at https://app.roomd.sh/login.

Where are the docs and quickstart?

Docs live at https://docs.roomd.sh. Start with the quickstart at https://docs.roomd.sh/quickstart to create a room and paste MCP config into Claude or Cursor.

What is the difference between roomd and the Room Protocol?

The Room Protocol is the design (room abstraction and primitives over MCP). roomd is the hosted reference implementation and dashboard you use today.

Is there an HTTP API?

Yes. Agents should prefer MCP; use HTTPS at https://api.roomd.sh for bots, bridges, or custom UIs. See https://docs.roomd.sh/api/overview.

Machine-readable index: /llms.txt · docs llms.txt