Python Tool (Canonical v10)

The python tool runs Python code inside a canonical step pipeline (step.tool).

Python task modes (implementation-defined, common in NoETL runtimes):

• Pure code mode: set a top-level result = ... in your code
• Legacy mode: define main(...) and return a JSON-serializable object

Canonical reminders:

• Use workload for immutable inputs, ctx for execution-scoped state, iter for iteration-scoped state.
• Use task.spec.policy.rules for retry/fail/jump/break/continue.

---

Basic usage (pure code mode)

- step: transform
  tool:
    - run:
        kind: python
        args:
          items: "{{ workload.items }}"
        code: |
          result = {"count": len(items)}
        spec:
          policy:
            rules:
              - when: "{{ outcome.status == 'error' }}"
                then: { do: fail }
              - else:
                  then: { do: break }

---

External scripts (script descriptor)

Python also supports the canonical script descriptor (uri + source) to load code from GCS/S3/HTTP/filesystem.

- run_external:
    kind: python
    script:
      uri: gs://my-bucket/scripts/analyze.py
      source:
        type: gcs
        auth: gcp_service_account
    args:
      dataset: "{{ workload.dataset }}"

See documentation/docs/reference/script_execution_v2.md for the script descriptor.

---

See also

• Variables/scopes: documentation/docs/reference/variables_v2.md
• Retry semantics: documentation/docs/reference/retry_mechanism_v2.md
• Result storage (reference-first): documentation/docs/reference/result_storage_canonical_v10.md