An agent member is a member of your workspace that is an agent, not a person. Use one for an agent that works with nobody at the keyboard: a cloud agent, a CI or scheduled job, or a third-party agent connected over MCP or the API. It tracks its own time, under its own name, and that time appears on the Team list, timesheets, reports and invoices like anyone else's, labeled Agent.
This is different from agent time, which records a coding agent's work as time belonging to the person who ran it. An agent member's time is its own.
Owners and admins add agent members under Settings → Agent members (or + Add agent member on the Team page). Turn on Agent members under Settings → Add-ons first.
Give it a name, a role (member, manager or admin), and optionally a billable rate, a cost rate and the manager who approves its timesheets. When you add it you are shown its token, once. Copy it then: it cannot be shown again. Give the token to the agent as Authorization: Bearer <token> on /api/v1, or use it to connect the agent to the MCP server at /api/mcp.
POST /api/v1/time_entries or the MCP log_time tool, and start and stop timers.An agent member's token can write over MCP, because it is the agent's only credential. A person's personal token stays read-only there.
A person's entries are unchanged. An agent member's entries are held to sanity limits, and anything outside them is refused with a reason and a machine-readable code:
started_at and ended_at), not just hours;/track linking.The assistant can list agent members, add one, rotate its token and disable it (owners and admins), and approve or return weeks. It never sends email for any of these.
Updated 2026-11-03. Still stuck? Contact support or return to the Help Center.