One Headless Work, from input to result

This reader displays selected fields from a retained execution record. It does not run commands or connect to a service. It is not a terminal screenshot or a live session.

The product last reported human acceptance pending. The later approval of the local exercise and guide was recorded separately; the product state was not updated.

Step 1 / 6

Check the service

What the operator did

The operator queried the intended local service with status and health. The health command returned exit 1.

Open this step’s command reference

What the product returned

Actual response excerpt · other fields omitted

{
  "service_state": "RUNNING",
  "liveness": {
    "reason_code": null,
    "status": "READY"
  },
  "readiness": {
    "reason_code": "AUTHORITY_READINESS_OBSERVATION_UNAVAILABLE",
    "status": "UNKNOWN"
  },
  "runtime": {
    "reason_code": "LIVE_RUNTIME_OBSERVATION_UNAVAILABLE",
    "status": "UNKNOWN"
  }
}

How to read it

RUNNING and READY describe service presence and liveness. Readiness and runtime observations remained UNKNOWN; this was not an authorization to execute Work.

Why the next step matters

This local exercise had its own approval, and the normal operation admission path evaluated submission. Do not treat an UNKNOWN prerequisite on another installation as permission to proceed.

Step 2 / 6

Read the rejected input

What the operator did

The first intent was “Create a local lineage artifact for this approved operator exercise”. It omitted the required audit term. This is the failed historical input, not a command to try.

Open this step’s command reference

What the product returned

Actual response excerpt · other fields omitted

{
  "error": {
    "code": "PACKR_OPERATOR_STATUS_UNAVAILABLE",
    "message": "protocol failure: PACKR_IPC_MALFORMED_REQUEST: invalid request: SEMANTIC_INTENT_TYPE_UNRESOLVED"
  }
}

How to read it

The outer error name alone does not establish a service outage. The nested reason identifies SEMANTIC_INTENT_TYPE_UNRESOLVED at preparation.

Why the next step matters

Before correcting and submitting anything, the operator checked whether the original attempt had produced a Work.

Step 3 / 6

Check whether Work exists

What the operator did

The operator read work list using the same installation and account. These collections were empty in that response.

Open this step’s command reference

What the product returned

Actual response excerpt · other fields omitted

{
  "works": [],
  "admissions": [],
  "executions": [],
  "reports": [],
  "receipts": [],
  "closures": []
}

How to read it

The recorded preparation failure did not reach Work execution. Session metadata did persist, so the failed attempt was not a zero-change operation.

Why the next step matters

The operator verified the supported input before one corrected invocation. An empty scoped list is not a universal guarantee that no Work exists; uncertain completion requires reconciliation.

Step 4 / 6

Correct the intent once

What the operator did

The corrected intent was “create audit artifact”, submitted for CREATE_AUDIT_ARTIFACT. It created the Work below and the CLI returned exit 0.

Open this step’s command reference

What the product returned

Actual response excerpt · other fields omitted

{
  "result": {
    "work": {
      "work_id": "work-headless-c36fddb388d8599d9f5e70e40296eae13ccfa8ddaccf3976daacfa18e2ce373e",
      "state": "AWAITING_HUMAN_ACCEPTANCE"
    },
    "reports": [
      {
        "final_result": "AUTOMATION_COMPLETED_HUMAN_ACCEPTANCE_PENDING"
      }
    ],
    "replayed": false,
    "execution_retry_allowed": false,
    "human_acceptance_claimed": false
  }
}

How to read it

Automation completed, with human acceptance still pending. The response says replayed is false and execution retry is not allowed. Read the Work outcome as well as the command exit.

Why the next step matters

Save the exact returned work_id. Use that identity for list, show and report review; do not submit the operation again.

Step 5 / 6

Follow the same Work

What the operator did

The operator matched the returned Work ID across submission, list, show and the selected report. This is the report excerpt for that same Work.

Open this step’s command reference

What the product returned

Actual response excerpt · other fields omitted

{
  "work_id": "work-headless-c36fddb388d8599d9f5e70e40296eae13ccfa8ddaccf3976daacfa18e2ce373e",
  "execution_count": 1,
  "final_result": "AUTOMATION_COMPLETED_HUMAN_ACCEPTANCE_PENDING",
  "validation_results": "PASSED",
  "closure": {
    "disposition": "AWAITING_HUMAN_ACCEPTANCE",
    "human_acceptance": null,
    "human_acceptance_required": true
  }
}

How to read it

There was one actual execution and canonical validation PASSED. The closure still required human acceptance. Validation and final human acceptance are separate states.

Why the next step matters

Inspect the actual report files before claiming that the output has been delivered and reviewed.

Step 6 / 6

Verify the report files

What the operator did

The operator opened the returned report locations, compared the physical JSON and PDF file hashes with these projection fields, and viewed the PDF in a native viewer.

Open this step’s command reference

What the product returned

Actual response excerpt · other fields omitted

{
  "json_file_sha256": "d7c014b72ebb839269c0ab11c0c23bea7e07416178ec31583b7d10b6bc0ef996",
  "pdf_sha256": "sha256:1265657f5b51e8e63dd86a3977a91731f4f6893b37e1f1431a6977d88ddbb4f3",
  "report_sha256": "sha256:f97ee5c9726fd67e47343eb7425e56af11ad74f56a77e38636311a335ec0fc48"
}

How to read it

The file hashes matched. report_sha256 is a separate internal report digest; it is not the raw JSON or PDF file checksum. Your own run will have its own values.

Why the next step matters

The recorded result is automated completion with human acceptance pending. Retain your Work ID, outcome and approved report handoff; raw reports remain private.