Use Agent from Claude, ChatGPT or Codex
Add Agent to Claude, ChatGPT, Claude Code or Codex. The model runs there on your plan and uses Agent's tools to read and assign work.
ConnectLink to Connect
- Claude (web, desktop, mobile): Settings, Connectors, Add custom connector. Enter your Agent server's
/mcpaddress. The connector address is on Connections. - Claude Code:
claude mcp add --transport http agent <address>, then/mcpto sign in. There is also an Agent plugin for Claude Code. - ChatGPT: add an app with the same address, in developer mode or as a workspace custom app.
- Codex: install the Agent plugin, then
codex mcp login agent.
Agent opens a consent page. Pick the organization, keep or clear each permission and select Allow.
PermissionsLink to Permissions
- Read work: workers, work items and their sessions.
- Assign work and post messages: needs the member role.
- Read work context: briefs, checks, evidence, knowledge, specs and the decision log.
- Search memory.
- See pending approvals. Deciding them stays in Agent.
- Work on your own plan: take work Agent hands to your Claude or ChatGPT, and report results. Needs the member role.
- Stay connected.
The tool acts as you, within your role. What it posts is marked as coming from the connection. Nothing in the connector approves an action or writes memory.
RevokeLink to Revoke
Select Revoke on Connections. An admin can turn connections off for the whole organization there; every connection then stops at once.
Checked against the product on 2026-10-05.
Was this helpful?
Related articles
- Background runs on your own planAgent can start work on your own Claude routine or ChatGPT Workspace Agent while you are away. You save its trigger in Agent.
- Subscriptions and vendor termsYour Claude and ChatGPT plans work through the runner on your own computer. Your plan's terms still apply, so an admin accepts a caveat before runners are allowed.
- API tokensAPI tokens are machine credentials for CI, scripts and MCP clients. Each token has one role and expires.