Skills
Navigate
skills/davila7/claude-code-templates/security-ownership-map/Gen Agent Trust Hub

security-ownership-map

Pass

Audited by Gen Agent Trust Hub on Oct 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes git commands using the subprocess module to retrieve repository history from local paths.\n
  • Evidence: subprocess.Popen calls in scripts/build_ownership_map.py and scripts/community_maintainers.py execute git log to extract author and file metadata.\n
  • Evidence: scripts/run_ownership_map.py utilizes subprocess.run to orchestrate internal Python script execution.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted metadata from git logs which could potentially contain adversarial instructions targeting the AI agent.\n
  • Ingestion points: git log metadata (author names, emails, and file paths) is parsed in scripts/build_ownership_map.py and scripts/community_maintainers.py.\n
  • Boundary markers: Data is delimited within structured CSV files (people.csv, files.csv) and JSON summaries (summary.json).\n
  • Capability inventory: The skill possesses the capability to read local file systems (via git history) and write analysis results to local directories.\n
  • Sanitization: No explicit sanitization or filtering is performed on the ingested git metadata to prevent malicious instruction patterns from entering the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 5, 2026, 10:48 PM