agents-md-sync

Pass

Audited by Gen Agent Trust Hub on Aug 19, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a bundled local shell script (scan-repo.sh) to perform static analysis of the target repository's structure, language distribution, and documentation drift. The script relies on standard system utilities (find, git, awk) and performs only local, non-destructive read operations.
  • [SAFE]: The skill implements a specific security feature called 'Global Config Repo Protection.' It identifies if the target repository contains global agent instructions (e.g., in ~/.claude or ~/.codex) and automatically halts execution to prevent the AI from poisoning its own system-level instructions.
  • [SAFE]: The skill processes codebase content to generate summaries, which is a standard documentation functionality. It mitigates potential indirect injection risks from source code comments by using a structured multi-agent workflow where a separate 'review sub-agent' verifies all generated content against strict quality criteria before the final files are written.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 19, 2026, 05:52 AM
Security Audit — agent-trust-hub — agents-md-sync