metrics-anomaly-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external metrics, logs, and observability data.
- Ingestion points: Processes user-supplied 'metric series', 'alert rules', 'deployment events', and 'monitoring evidence' as defined in
prompts/metrics-anomaly-analysis.md. - Boundary markers: The skill instructions explicitly demand that the agent 'Audit input completeness, credibility, recency, and comparability' and 'separate confirmed facts, working assumptions, and open questions'.
- Capability inventory: The skill is purely analytical and document-based; it does not include any scripts, network operations, or file-write capabilities.
- Sanitization: The instructions require the agent to 'Never invent system behavior... absent from the evidence' and to mark unsupported claims as hypotheses, reducing the risk of the model being misled by malicious patterns in the input data.
Audit Metadata