The Agent Skills Directory

[SAFE]: The skill implements a strictly defined, multi-phase review process for project governance documents. It focuses on auditing "Guardrails" for executability, evidence-based standards, and impact on downstream workflows.
[SAFE]: The execution uses TodoWrite for transparent progress tracking and follows a rigid severity grading (P0/P1/P2) for findings, which enhances safety by ensuring all critical issues must be resolved (P0=0).
[SAFE]: Data handling is restricted to reading local project documentation (Guardrails) and providing feedback via report templates. There are no external network calls, sensitive file access (e.g., credentials or SSH keys), or dynamic code execution.
[SAFE]: The AskUserQuestion module is used solely for clarifying project scope, action types, and conflict resolution, following standard interaction patterns without attempting to exfiltrate data.
[SAFE]: All references and templates (e.g., references/report-templates.md, references/askuser-templates.md) are project-local and contain legitimate instructional content relevant to the skill's purpose.

guardrails-reviewer