anti-pattern-detection

Run the eight-pattern detection pipeline against your capabilities and requirements. It checks for solution masquerading, missing EARS acceptance criteria, vague actors, subjective terms, missing triggers, non-testable outcomes, oversized capabilities, and passive voice, then reports each violation with a suggested fix.

The Anti-Pattern Score equals (8 minus violations) divided by 8. Scores below 0.70 block the capability, scores between 0.70 and 0.85 trigger warnings with fixes, and scores above 0.85 pass. Each CRITICAL finding also reduces agent confidence by 0.10 and each HIGH finding by 0.05.

Three patterns are CRITICAL: solution masquerading as a requirement (technology terms in a capability), missing acceptance criteria (no EARS patterns found), and mega-capabilities (more than 10 functional requirements, 8 criteria, or 7 entities). Vague actors, subjective terms, and non-testable outcomes are HIGH severity.

Capabilities scoring below 0.70 are blocked from finalization. The output lists each violation with its pattern name, severity, original text, issue explanation, suggested fix, and required action, plus a summary of violations by severity and the overall quality score.

Run it as a quality gate before finalizing any capability or requirements specification. It works best after drafting requirements but before decomposition or implementation planning, since mega-capability findings can trigger a separate decomposition step.