Skill Comply -- Automated Compliance Measurement

Measures whether coding agents actually follow skills, rules, or agent definitions.

When to Use

• After adding new rules or skills, to verify agent compliance
• Periodically as part of quality maintenance
• When asking "is this rule actually being followed?"

How It Works

1. Auto-generate specs: Extract expected behavioral sequences from any .md file
2. Auto-generate scenarios: Create test prompts with decreasing strictness (supportive -> neutral -> competing)
3. Run agent: Execute claude -p and capture tool call traces via stream-json
4. Classify: Match tool calls against spec steps using LLM (not regex)
5. Check ordering: Verify temporal ordering deterministically
6. Report: Generate self-contained reports with spec, prompts, and timelines