Starts the policy runtime locally and exercises API scenarios with reproducible sample requests, including allow, deny, approval, budget, malformed input, and audit checks. Use when the user asks to deploy/run locally, smoke test endpoints, or validate runtime behavior through API calls.
Run the app locally and validate runtime branches via API calls.
Apply this skill when the user asks to:
Run from repository root.
Task Progress:
python -m pip install -e ".[dev]"Use either:
python scripts/run_api.pyuvicorn app.api.main:app --host 127.0.0.1 --port 8000Health check:
curl.exe http://127.0.0.1:8000/docsUse the payloads and commands in scenarios.md.
Always include:
workflow_type: "invoice_governance"request.input_payloadrequest.actorrequest.budgetFor each scenario:
run.run_idrun.status, run.current_state, run.pending_tool_id, run.stepsGET /workflows/{run_id}/auditAlso inspect:
artifacts/audit.jsonlartifacts/runs.jsonlPOST /workflows/executeGET /workflows/{run_id}POST /workflows/{run_id}/approvalPOST /workflows/{run_id}/resumeGET /workflows/{run_id}/auditCOMPLETEDBLOCKEDAWAITING_APPROVAL then COMPLETED after approval + resumeBLOCKEDBLOCKEDFAILEDrun.steps[-1].error when present.