Compliance

Overview

Compliance is the final quality gate before a task is archived. It detects the task type automatically and applies the appropriate verification checklist. Whether the work is code, documentation, research, legal, content, or infrastructure, Compliance ensures nothing ships without meeting all stated requirements.

Supported Task Types

• Software — 7-step checklist: PRD alignment, code simplification, dead code detection, test coverage, linter runs, test execution, and CI/CD impact analysis.
• Documentation — completeness, accuracy, consistency, cross-references, audience appropriateness.
• Research — methodology validation, citation completeness, argument coherence, scope compliance.
• Legal — jurisdictional compliance, definitions consistency, structural integrity, rights and obligations clarity.
• Content — factual accuracy via the factcheck skill, AI pattern removal via humanize, platform requirements, editorial standards.
• Infrastructure — configuration accuracy, rollback plan, monitoring setup, security posture, discovery probe verification (verify assumptions against live APIs during /dr-prd, not at /dr-do).

When It's Used

The compliance agent runs this skill during the /dr-compliance command. For Level 3-4 tasks, it is mandatory before archiving. For Level 1-2, it is optional. Mixed tasks (e.g., code + docs) receive checklists from each relevant type.

Output

A compliance report with per-step results (PASS / FAIL / N/A), an overall verdict (COMPLIANT, NON-COMPLIANT, or COMPLIANT_WITH_NOTES), and a list of remaining risks. Reports are saved to datarim/reports/.