/schedule: 'Run now' fails on both CLI (RemoteTrigger run action) and web UI

Resolved 💬 3 comments Opened Apr 28, 2026 by charis-algomo Closed May 2, 2026

Summary

Triggering an on-demand run of an existing routine fails through both the RemoteTrigger tool (action: "run") and the web UI's Run now button. The routine itself is healthy and continues to fire on its cron schedule — only the manual-run path is broken.

Reproduction

  1. Create any routine via /schedule (e.g., a recurring agent on cron_expression: "15 */2 * * *")
  2. Either:

Observed

CLI path — HTTP 400:

{"type":"error","error":{"type":"invalid_request_error","message":"trigger_id: Extra inputs are not permitted"}}

Tried with no body, with body: {}, and with various body shapes — all fail identically. get/update/create against the same trigger_id all succeed, so the routine ID is valid.

Web UI path — orange toast: "Failed to start run. You can try again." Retrying produces the same toast.

Expected

Either path should enqueue a one-off run of the routine.

Suspected cause

The validation error message trigger_id: Extra inputs are not permitted suggests the request body is being validated with a strict (extra=forbid) schema and trigger_id — which belongs only in the URL path POST /v1/code/triggers/{trigger_id}/run — is being inserted into the body as well, presumably by a shared serializer also used by get/update. Same root cause likely behind the web UI's silent toast.

Workaround

Wait for the next cron fire — the cron path is unaffected.

View original on GitHub ↗

This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗