reviewing-claude-config

Security scanning for Claude configs detects critical issues like committed settings.local.json, hardcoded secrets, overly broad permissions, and dangerous auto-approvals. This Skill applies specialized checklists to CLAUDE.md, skills, agents, prompts, and settings files, identifying risks before deployment and delivering actionable feedback.

The Skill reviews CLAUDE.md, SKILL.md, agent configurations, prompts, commands, and settings within .claude directories. It validates YAML frontmatter, enforces progressive disclosure (500-line limits), checks file reference integrity, and ensures prompt engineering best practices across all configuration types.

Yes. Use this Skill to automatically scan new Claude Skill changes for security risks, structural compliance, and prompt quality. It routes to specialized review checklists based on file type, validates YAML structure, enforces token-efficient guidelines, and reports issues before you commit.

Automated review detects hardcoded secrets, committed settings.local.json files, dangerous command auto-approvals, overly broad permissions, and other critical vulnerabilities. It applies security-first validation through specialized checklists tailored to each configuration file type.

The Skill identifies the configuration file type (skills, agents, CLAUDE.md, prompts, settings) and applies a token-efficient, specialized review checklist for that type. This ensures relevant security and quality checks without unnecessary overhead.

The Skill enforces progressive disclosure guidelines, such as the 500-line limit for main SKILL.md files. It flags oversized sections and provides feedback to improve token efficiency and readability before deployment.