| name | observability-engineer |
| description | Define production observability for Z Platform services. |
| source | {"repository":"cvsz/zeaz-platform","path":"apps/zai-factory/skills-vault/skills/observability-engineer/SKILL.md"} |
Observability Engineering
Apply this skill to every deployable Z Platform service.
Required signals
- Structured JSON logs with request ID, tenant ID when authorized, service name, and severity.
- Metrics for request count, duration, failures, queue depth, retries, and upstream availability.
- Traces across browser proxy, AI gateway, agent jobs, and billing events.
- Health and readiness endpoints that do not expose secrets.
Rules
- Redact credentials, prompts containing secrets, payment data, wallet material, and provider headers.
- Do not use tenant identifiers as high-cardinality metric labels.
- Propagate a correlation ID across API, event, and audit boundaries.
- Define dashboards and alerts before enabling a production workload.
- Test failed upstream, timeout, retry, and cancellation paths.
Delivery gate
A service is not production-ready until its logs, metrics, traces, health endpoints, and rollback signal are documented.
