security-ownership-map

Pass

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill invokes the git CLI to retrieve repository commit history. This is implemented securely using list-based argument passing in subprocess calls, which mitigates the risk of shell injection.
  • [EXTERNAL_DOWNLOADS]: The skill requires the networkx Python library for computing graph metrics and community detection. This is a well-known, trusted package in the data science ecosystem.
  • [PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it processes untrusted content from git repositories (such as author names and commit metadata).
  • Ingestion points: scripts/build_ownership_map.py and scripts/community_maintainers.py parse data from git log output.
  • Boundary markers: The generated CSV and JSON artifacts do not include delimiters or specific instructions for an LLM to ignore potentially malicious embedded content.
  • Capability inventory: The skill performs local file system write operations to generate its reports.
  • Sanitization: Commit metadata is processed without sanitization or validation, allowing strings from a repository to reach the final analysis output directly.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 14, 2026, 08:24 PM