# get_job_status (/mcp/get-job-status)



Checks a job from [`create_score`](/mcp/create-score) or [`calibrate`](/mcp/calibrate), as [`GET /v1/jobs/{job_id}`](/reference/get-job-status). Read-only.

| Input     | Description                                  |
| --------- | -------------------------------------------- |
| `job_id`  | Required. From `create_score` or `calibrate` |
| `api_key` | `/mcp/keys` only                             |

Returns `job_id`, `complete` and, if the job failed, `error` with a `type` and a `message` saying what to do. Call it about every 30 seconds until `complete` is `true`, not in a tight loop: creating a score takes about 5 to 10 minutes, and there's no progress to report.
