What problem does it solve? AI agents produce more code than anyone reviews, and without a written quality bar they quietly lower standards — adding suppressions, skipping tests, or weakening thresholds to reach green. This Skill turns a project's quality expectations into a versioned, machine-checkable contract that outlives any single session. ## Core Features & Use Cases - Constraint Interview & Defaults: Runs a four-question intake with sane defaults, then writes a CONSTRAINTS.md covering the floor rules, enforced dimensions with numbers, measured ratchets, and tracked exceptions. - Tool Wiring & Lifecycle Placement: Maps each dimension (types, lint, coverage, security, performance, accessibility) to a de facto tool like Semgrep, gitleaks, osv-scanner, Lighthouse, or axe, and assigns checks to fast, task, and CI stages. - Floor Guard: Ships a reference implementation that scans the git diff for the five bar-lowering moves — lowered thresholds, weakened tests, new suppressions, unfinished stubs, and surprise exceptions. - Use Case: Before running an autonomous build loop on a TypeScript repo, use this Skill to record an 80% changed-line coverage gate, block new eslint-disable comments, and catch any agent attempt to relax the bar in review. ## Quick Start Ask the agent to set up project constraints and define our quality standards in CONSTRAINTS.md for this repository.
