Argus

AI instruction file reviewer for GitHub.

Automatically reviews CLAUDE.md, AGENTS.md, .cursorrules, and 9 more file patterns. Scores against the Argus Governance Standard v1.1 -- 13 checks, document-type-aware scoring, zero false positives.

Install on GitHub

Supported frameworks

• Claude Code (CLAUDE.md)
• AGENTS.md (open standard)
• GitHub Copilot (copilot-instructions.md)
• Cursor (.cursorrules, .cursor/rules/)
• Gemini CLI (GEMINI.md)
• Windsurf (.windsurfrules)

What it checks

• Silent inference patterns
• Authority boundaries
• Scope limitations
• Audit trail instructions
• Error handling
• Output format specification
• Identity definition
• Objective clarity
• Escalation paths
• Data handling guidelines

Links

Source code & documentation

Scoring methodology