Result Verification

Result verification

Use verify when a command's exit status is insufficient to establish that its output is usable. A verifier can check file contents, query a database, or confirm that an uploaded object exists.

jobs:
  - name: export
    command: python export.py
    schedule: "0 * * * *"
    verify:
      command: python check_export.py
      timeout: 30

The verifier runs after the command passes its failsWhen rules. It inherits the job's environment, working directory, shell, privileges, secrets, and job API context. command accepts a shell string or an argument list. timeout defaults to 60 seconds and must be finite and positive. Set verify: null to clear an inherited verifier.

The verifier must exit with code zero. A nonzero exit, timeout, or launch failure fails the whole run, triggers its failure reporters, and follows its retry policy. Retrying runs the command and verifier again. A failed or cancelled main command skips verification. Cancellation during verification terminates the verifier. Its timeout is separate from executionTimeout.

The command's exit code remains intact. History includes a separate verification object with the check's outcome, exit code, timing, and diagnostics. Diagnostic text is redacted, retains at most 64 lines per stream, and is capped at 16384 characters per stream; oversized lines are discarded. output_truncated identifies truncation of retained output. Live output uses the verify.stdout and verify.stderr stream names. Report templates can inspect verification.

Executable DAG tasks accept the same verify block. Downstream tasks wait for verification to pass, and the task detail shows its result. Approval gates reject verify. A resource pool claim remains occupied through verification.