inspect-infrastructure-health
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFE
Full Analysis
- [Strict Input Validation]: The skill implements a robust safety contract that explicitly forbids the construction or acceptance of raw inputs such as PromQL, SQL, URLs, or headers. This effectively mitigates risks associated with injection attacks targeting the underlying data source.
- [Least Privilege Tool Usage]: Access is restricted to a specific set of predefined tools (
grafana_profile_status,grafana_run_pack, etc.), and the skill is prohibited from writing to any environment, datasource, or deployment, ensuring its operations remain read-only. - [Opaque Reference Handling]: The skill uses opaque references for report and evidence data rather than direct URLs. This pattern is a security best practice that prevents the agent from interacting with arbitrary or malicious external links.
- [Explicit Environment Guardrails]: The workflow requires confirmation before using live environments and mandates standardized status interpretations (COMPLETE, STALE, ERROR), which reduces the risk of the agent providing misleading or unverified health claims.
Audit Metadata