Client common failures

Use this page when a local workspace cannot connect to Team OS or open the expected Team resources.

Symptom	What to check
`--api-url must start with http:// or https://`	Use the full server URL, including `https://`.
`login failed (401)` or `unauthorized`	Check the email, password, invitation state, and server URL.
Connected, but no Team is available	A Company Admin can sign in without Team access. Request access or ask for a direct Team role.
Expected Team is missing	Confirm the Team is active and the user still has effective access through Company ownership, an active Company grant, or a direct Team role.
`No clients are available for this account.`	Confirm the selected Team has active clients. For a Team Member, check client grants.
Saved login expired	Run `npm run team -- login` again.
Command Centre shows signed out after terminal login	Start both with the same `AGENTIC_OS_TEAM_CONFIG_DIR`.
New Team work is blocked while the server is unavailable	Restore the connection. Command Centre does not use stale Team or Solo context as a fallback.
Existing Team reply cannot continue	Restore access or the service. Conversation-only mode also needs the saved agent session.
Health check fails	Ask the server operator to check `/v1/health` and the API deploy logs.

What the server decides

The local profile identifies a server and signed-in user. The server decides which Teams, clients, skills, memory, and files that user can access.

The local workspace cannot grant itself a Team, role, client, skill, or user identity by sending different IDs in a request.

Switching Teams affects new work. It does not move an existing task or conversation to a different Team.