Troubleshooting
Agent won't start
vibe status # what's the agent doing?
vibe health # HTTP probe
Common issues:
- Port 3005 in use — agent auto-picks the next free port, but if it can't, set
PORT=<n>env var - tmux/ttyd/cloudflared missing — run
vibe setupto auto-install system deps - Permission errors on
~/.boff/vibecontrols/—chmod -R u+rw ~/.boff/vibecontrols
Agent shows as Disconnected in the web app
The agent is running but the backend can't see it.
vibe status— is the tunnel up?- Check the agent's outbound tunnel URL:
curl -H "x-agent-api-key: $KEY" http://localhost:3005/api/agent/tunnel - Try
vibe restart(or kill +vibe start) - Re-push gateway credentials from the web app: Targets → <target> → Re-authenticate
Session "Terminal started but no tunnel URL was returned"
The agent stored a stale OAuth credential. The agent should self-heal — but if not:
vibe restart
If still failing, re-push gateway auth from the web app.
Session won't start
vibecontrols sessions describe <session-id>
Look at the status and lastOutput fields. Common causes:
workingDirectorydoesn't exist on the agent's machinecommandnot found on PATH- Session-type plugin not loaded (e.g., wezterm session but wezterm not installed)
Tunnel returns 502 / hangs
- Is the underlying service actually listening on the port?
vibecontrols sessions create --type TERMINAL→curl localhost:3000on the agent vibecontrols tunnels list— is the tunnel still active?- For Cloudflare quick tunnels: trycloudflare.com has rate limits; spin up a new tunnel
- For VibeTunnels: check your
frpsis up
CLI says "unauthorized"
vibecontrols auth status
vibecontrols auth login # re-auth if token expired
vibecontrols workspace set <id> # if you switched workspaces
"Failed to load vibe"
Usually a backend/DB schema drift after an upgrade. Wait a few minutes (migrations may still be running) or contact support.
Reset everything
vibe autostart uninstall
rm -rf ~/.boff/vibecontrols
npm uninstall -g @vibecontrols/agent @vibecontrols/cli
Then re-install fresh.