API reference

Outcome Feedback API

The difference between a registry and a readiness network. An HTTP 200 is not task success — agents report whether the result was actually usable, and scores recalibrate on it.

POST/v1/capabilities/outcome

Outcome fields

Field	Description
call_completed	The invocation finished without transport-level failure.
result_usable	The agent could actually use the result for the task.
subgoal_completed	The user-visible subtask was completed.
fallback_needed	The agent had to switch to a fallback capability.
user_repair_needed	A human had to fix the result afterwards.
agent_judgement	Self-assessed score, confidence, and structured reason codes.

request.sh

curl -X POST http://101.32.68.30/v1/capabilities/outcome \
  -H 'Authorization: Bearer $API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{
    "trace_id": "tr_anon_123",
    "capability_id": "cap_pdf_extract_table",
    "action": "extract_table",
    "agent_runtime": "claude-code",
    "agent_version": "1.0.0",
    "task_family": "pdf_table_to_csv",
    "outcome": {
      "call_completed": true,
      "result_usable": true,
      "subgoal_completed": true,
      "fallback_needed": false,
      "user_repair_needed": false
    },
    "agent_judgement": {
      "score": 0.86,
      "confidence": 0.74,
      "reason_codes": ["schema_valid", "result_used_in_final_answer"]
    },
    "privacy": {
      "contains_raw_prompt": false,
      "contains_user_file": false,
      "contains_pii": false
    }
  }'

Privacy contract

Outcome reports are structurally validated: payloads flagged as containing raw prompts, user files, or PII are rejected with 422. Reports are de-identified on ingestion and weighted by each agent's historical reliability.

response.json

{
  "accepted": true,
  "outcome_id": "out_123",
  "score_refresh_scheduled": true
}