What problem does it solve? AI coding agents produce more code than anyone reviews, and without a written quality bar they take the cheapest road to green: silencing linters, skipping tests, lowering thresholds, and leaving stubs. This Skill turns your project's standards into a written, mechanically checkable contract that outlives the session. ## Core Features & Use Cases - Constraint Interview with Defaults: Asks four focused questions about which dimensions matter (coverage, security, performance, accessibility, architecture), supplying sane default thresholds when you have no number in mind. - CONSTRAINTS.md Contract: Writes a repo-root file recording the always-enforced floor (no new suppressions, no skipped tests, no stubs, no secrets), numbered enforced dimensions with the exact command that checks each, measured-only ratchets, and exceptions with owners and expiry dates. - Diff-Scoped Floor Guard: Ships a reference implementation that watches the diff for the five bar-lowering moves — weakened thresholds, easier tests, silenced checkers, unfinished work, and surprise exceptions — with exit codes 0/1/2. - Use Case: Before running an autonomous build loop on a TypeScript project, use this Skill to record an 80% changed-line coverage gate, gitleaks and Semgrep security checks, and a Lighthouse LCP budget, then wire them into fast/task/CI stages so the agent fixes its own output. ## Quick Start Ask the agent to set up quality constraints for this project and write them into CONSTRAINTS.md with enforced coverage and security checks.
