The Agent Skills Directory

Data Exposure & Exfiltration (LOW): The skill performs network operations and accesses system memory metrics, which are standard for health-check functionality but require monitoring.\n- Evidence: templates/health-check.ts implements createHttpCheck which uses the fetch() API to query external URLs.\n- Evidence: templates/health-check.ts implements createMemoryCheck which accesses process.memoryUsage() to gather system-level telemetry.\n- Indirect Prompt Injection (LOW): The skill ingests data from external network responses and database error messages which could serve as an injection vector if health reports are directly interpreted by an LLM.\n- Ingestion points: Network responses in createHttpCheck and error messages in createDatabaseCheck (file: templates/health-check.ts).\n- Boundary markers: Absent; the templates do not include delimiters or specific instructions for an agent to ignore embedded content.\n- Capability inventory: Network requests via fetch() in templates/health-check.ts and templates/circuit-breaker.ts.\n- Sanitization: Absent; error messages and response status data are passed directly into the health report objects without filtering.

reliability-engineering