Troubleshooting

Diagnose stale dashboards, missing agent binaries, config mismatches, and session-runtime issues.

Most failures are operational, not conceptual.

​

Common failure modes

• dashboard state looks wrong after config edits
• the running process cannot see agent binaries in PATH
• the browser and API are hitting different origins
• a repo board is being watched twice through canonical and symlinked paths

​

Useful commands

co doctor
co status
curl -s http://localhost:4747/api/config
curl -s http://localhost:4747/api/agents

​

Recovery checklist

1. Restart the process with the intended workspace and CO_CONFIG_PATH.
2. Confirm the agent binary is visible to the process that launches Conductor.
3. Check the browser origin and API origin match.
4. Inspect config and agent endpoints before assuming the board parser is wrong.