pre-work-research
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill employs standard read-only commands including ls, cat, find, and grep to analyze the repository structure and content, which is consistent with its purpose.
- [SAFE]: External activity is limited to package information lookups and documentation searches, with no evidence of remote code execution or exfiltration to untrusted domains.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from repository files via cat and grep (Ingestion points). No specific boundary markers are used to isolate file content (Boundary markers). It possesses capabilities to post comments to GitHub issues via the gh CLI (Capability inventory). No sanitization of the analyzed content is performed (Sanitization). Given the research use case, this surface is considered safe.
Audit Metadata