Codevisor Docs

Operations and troubleshooting

Monitor, update, restart, and troubleshoot the server.

Service status

codevisor status
codevisor logs --follow

These commands select the systemd user unit, system unit, or background-process log as appropriate. Use codevisor status --json for machine-readable server identity and harness readiness.

Health and identity

GET /v1/health reports the release, database state, process identity, and migration progress when present. GET /v1/info reports the server and machine IDs, kind, platform, architecture, hostname, bind address, and feature markers.

Updates

Standalone release builds check the stable release manifest. GET /v1/update reports update state; POST /v1/update/apply downloads the matching runtime and restarts when no prompt is in flight. A busy server rejects the restart without interrupting active work.

codevisor update always performs a fresh Stable check before applying an update. To include Alpha releases, run:

codevisor update --alpha

Common checks

  1. Confirm the process is listening on the expected address and port.
  2. Call /v1/health locally before diagnosing remote authentication.
  3. Verify the remote request includes Authorization: Bearer ….
  4. Check firewall, reverse-proxy, and TLS configuration independently.
  5. Confirm the service user's PATH contains the coding-agent CLIs you expect the server to use.
  6. Use POST /v1/harnesses/rescan after installing a CLI without restarting the server.
  7. Use codevisor plugin list and an explicit plugin restart when a pane returns 502 or a plugin is in the failed state.
  8. Inspect MCP connectionState and detail; OAuth servers may need authorization rather than a process restart.

On this page