What problem does it solve? Teams write skills, rules, and agent definitions for coding agents but have no way to verify the agent actually follows them. This Skill quantifies behavioral compliance by auto-generating test scenarios, executing agents, and reporting compliance rates with full tool call timelines. ## Core Features & Use Cases - Automated Spec Generation: Extracts expected behavioral sequences from any skill or rule Markdown file using an LLM, producing a YAML compliance spec with ordering constraints. - Three-Level Scenario Testing: Generates supportive, neutral, and competing prompts for the same task to measure whether a rule is followed even without prompt support. - LLM-Based Trace Classification: Classifies captured tool calls against spec steps by meaning rather than regex, then checks temporal ordering deterministically. - Use Case: After adding a new TDD rule to your agent configuration, run the tool against the rule file to see the compliance rate across three prompt strictness levels and identify which steps need hook promotion. ## Quick Start Run the compliance measurement on a rule file with the command: uv run python -m scripts.run path/to/rule.md, or add --dry-run to preview the generated spec and scenarios without execution cost.
