utility-pm-skill-auditor
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions specify executing local scripts via Bash, specifically
bash scripts/pre-tag-validate.sh. This is an intended function of a governance auditor to run repository-local validation tools. - [INDIRECT_PROMPT_INJECTION]: The skill processes repository files (PRDs, OKRs, Skill manifests) and re-derives counters from the filesystem.
- Ingestion points: The skill reads
CONTEXT.md,AGENTS.md,README.md, and various skill directories to compare declared vs. actual states. - Boundary markers: The skill does not explicitly define LLM boundary markers in its instructions, but relies on the
pm-skill-auditorsystem prompt atagents/pm-skill-auditor.mdfor execution context. - Capability inventory: The skill uses
bashfor validator execution and filesystem read operations. - Sanitization: No explicit sanitization of repository content is described before interpolation into the audit report.
Audit Metadata