agent-docs

Pass

Audited by Gen Agent Trust Hub on Jun 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: No malicious instructions, obfuscation, or sensitive data exfiltration patterns were detected. The skill is purely focused on documentation lifecycle management.\n- [COMMAND_EXECUTION]: The skill includes a shell script (evals/run-static-checks.sh) for development-time verification. It uses local repository paths to run static analysis tools, which is a standard practice for maintainable agent skills.\n- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface due to its core function of analyzing external repository documentation.\n
  • Ingestion points: Processes project files like AGENTS.md and llms.txt during audit and review workflows (found in SKILL.md).\n
  • Boundary markers: Uses structured templates (e.g., templates/audit-report.md) to isolate findings from original content.\n
  • Capability inventory: Writes output files to localized paths such as docs/audits/ (documented in SKILL.md).\n
  • Sanitization: Content is processed through analysis playbooks without explicit filtering of the analyzed text.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 26, 2026, 02:21 AM
Security Audit — agent-trust-hub — agent-docs