monitoring
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [SAFE]: The skill provides structured instructions for the AI agent to perform system discovery, metrics design, and alerting strategy. All identified patterns are standard operational guidelines for reliability engineering.
- [NO_CODE]: The skill consists exclusively of markdown documentation and YAML metadata. It does not contain any scripts, binary files, or automated commands that require execution within the host environment.
- [DATA_EXFILTRATION]: The skill explicitly includes safety instructions in Phase 5, Item 17, and Phase 4, Item 13, prohibiting the logging of sensitive data such as passwords, tokens, and PII, and requiring sanitization of database statements.
- [INDIRECT_PROMPT_INJECTION]: The skill establishes an ingestion surface for untrusted user data in Phase 1 (System Discovery) to generate monitoring strategies and queries.
- Ingestion points: System context and service names provided by the user in Phase 1.
- Boundary markers: Not explicitly defined in the prompt templates.
- Capability inventory: Generates monitoring configurations, PromQL, LogQL, and Terraform code (Phase 6, 10, 13).
- Sanitization: Instructions in Phase 4 and 5 mandate the exclusion of PII and the sanitization of specific telemetry attributes.
Audit Metadata