skill-audit
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes two local JavaScript scripts,
.agents/skills/platform/validate-skills.mjsand.agents/skills/platform/audit-semantics.mjs. These are intended for platform validation and semantic auditing of the skill bundle and its configuration. - [INDIRECT_PROMPT_INJECTION]: The skill processes other local files (skills, manifests, and lockfiles) as input. This creates an ingestion surface for indirect prompt injection if the files being audited contain malicious content; however, the skill is designed to produce a reporting artifact and does not execute the content of the skills it audits.
- [DATA_EXPOSURE]: The skill accesses repository-local metadata, including
.claude/links and lockfiles, to verify integrity. There are no network operations or exfiltration patterns observed.
Audit Metadata