Artifact Tool (Canonical v10)

The artifact tool loads (and optionally stores) externally stored results referenced by ResultRef/TempRef.

Use it when a prior step/task externalized a large output and you need the full body in a downstream step.

Canonical reminders:

• Prefer reference-first results for large payloads.
• Keep secrets out of event logs and out of ctx/iter.
• Handle control flow via task.spec.policy.rules and server routing via step.next.arcs[].

---

Basic usage (load)

- step: load_large_output
  tool:
    - load:
        kind: artifact
        action: get
        result_ref: "{{ ctx.some_result_ref }}"
        spec:
          policy:
            rules:
              - when: "{{ outcome.status == 'error' }}"
                then: { do: fail }
              - else:
                  then: { do: break }

You can also load directly by uri:

- load:
    kind: artifact
    action: get
    uri: "gs://my-bucket/results/run-123/output.json.gz"
    credential: gcp_service_account

---

Actions

action: get

Inputs (one of):

• result_ref (string or ResultRef/TempRef object)
• uri (string)

Optional:

• credential (string) for cloud URIs (S3/GCS)

Common URI schemes:

• noetl://... (ResultStore)
• nats-kv://...
• nats-obj://...
• s3://...
• gs://...
• eventlog://...
• file://... (or plain paths)

Outputs (tool result; wrapper depends on runtime):

• data (parsed JSON when possible, otherwise string)
• uri
• size_bytes
• compressed

action: put (optional)

Store data to a target uri and return metadata:

- store:
    kind: artifact
    action: put
    uri: "gs://my-bucket/exports/report.json.gz"
    credential: gcp_service_account
    compress: true
    data: "{{ ctx.report }}"

---

See also

• Result storage model: documentation/docs/reference/result_storage_canonical_v10.md
• ResultRef / TempRef: documentation/docs/reference/tempref_storage.md