Verify & Troubleshoot

Outcome

You can confirm production health quickly and isolate setup regressions.

Verification sequence

1. Health endpoint: GET /api/health returns status: "ok".
2. Slack ingress: mention bot and confirm thread reply appears.
3. Queue callback: verify callback route logs include successful processing.
4. Plugin auth (if enabled): run one real command and confirm expected result.

Symptom -> likely cause

Symptom	Likely cause	First check
Slack mention receives no reply	Missing webhook URL or token scopes	Slack app Event Subscriptions URL + scopes
Webhook 401/signature errors	Incorrect signing secret	SLACK_SIGNING_SECRET value
Message accepted but no threaded work	Queue callback trigger misconfigured	vercel.json trigger + callback route
Plugin commands fail auth	Missing credentials or OAuth state	Plugin env vars + /sentry auth flow

Useful signals

• webhook_handler_failed
• queue_callback_failed
• agent_turn_failed
• credential_issue_failed

Recovery order

1. Confirm most recent deploy/version boundary.
2. Validate env vars and webhook URLs.
3. Validate queue trigger and callback handler status.
4. Roll back to last known-good deployment if regression is immediate.

Next step

For runtime internals, read Execution Model.