Common issues
Resolve the most common setup, run, approval, export, and schedule problems.
The setup wizard does not finish#
- Confirm Base URL, Project Key / Slug, and API Token are populated.
- Select Test Connection and wait for a project name.
- Select at least one cycle.
- Keep Dry run selected while diagnosing.
- Try Finish Setup again.
Run Agent Now is unavailable#
The API can disable on-demand triggers with RUN_TRIGGER_ENABLED=false. Ask an operator to check /api/status and the deployment configuration. Do not bypass the UI with an unauthenticated request.
A run is blocked#
Open the run and read its blocked reason. Common causes include a suspended organization, confirmation requirements, live mode without an eligible plan, missing approved actions, or missing integration configuration.
An approval cannot be changed#
Approved and rejected decisions can become terminal. Refresh Approvals and inspect the current action state. If a decision is already terminal, start a new run rather than editing database state manually.
Exports fail#
- Confirm the session is still signed in.
- Confirm the run belongs to the active organization.
- Retry a CSV export before PDF to isolate rendering from data access.
- Record the HTTP status, run ID, format, and UTC time for support.
Schedule settings saved but no run appears#
The page stores preferences; it does not prove a durable scheduler is running. Verify GitHub Actions, cron, systemd, launchd, Railway, or another configured scheduler separately.
The dashboard says there is not enough history#
Complete and store one or more runs, then refresh. If standalone engine runs should appear, verify that their audit files are being ingested for the correct dashboard organization.