atelier-reviewlisted

Atelier suite — the adversarial pre-ship review. A full multi-step red-team of a FINISHED frontend build: fan out independent reviewers across dimensions (accessibility, performance, motion/reduced- motion, design-integrity/anti-slop, code quality), adversarially verify every finding to kill false positives, verify live in a real browser, synthesize a prioritized fix list, apply serially, and re-verify. Built to run with the ordinary Task/subagent tool — it does NOT require ultracode/Workflow. Use whenever a substantial build is "done" and needs a real audit before shipping, when asked to review / red-team / audit / "is this ready to ship" / find problems / pre-ship review a frontend. This is the heavyweight, adversarial counterpart to the atelier-perf-a11y self-checklist. Part of the Atelier suite.
luminary19/atelier · ★ 0 · Web & Frontend · score 68

Install: claude install-skill luminary19/atelier

# Atelier — Adversarial Review The builder is blind to their own work. This skill runs **independent, hostile reviewers** over a finished build, makes each finding **survive an adversarial refutation**, confirms everything **live in a real browser**, then applies fixes and re-checks. It is the structured, repeatable version of the multi-agent audit — **without** needing ultracode: it orchestrates ordinary subagents (the `Task` tool), which are always available. > **Inputs:** a built, runnable site/app + the **Direction Doc** (to check fidelity, via `atelier-direction`). > **Rubrics it reviews against** (cite cross-skill refs fully-qualified): `atelier-perf-a11y/references/ > preflight-checklist.md`, `…/anti-slop-preflight.md`, `…/accessibility.md`, `…/performance.md`, plus the > domain refs under `atelier-motion`, `atelier-scroll`, `atelier-webgl`, `atelier-typography`, > `atelier-layout`, `atelier-foundations`. Reviewer + verifier prompt templates and the findings contract: > **`references/reviewer-prompts.md`** (this skill's own ref). Live in-browser playbook: > **`references/live-verification.md`** (this skill's own ref). ## Where this sits — the quality ladder Three layers, escalating rigor — **same rubrics throughout**, just re-checked with more independence: 1. **perf-a11y self-checklist** (`atelier-perf-a11y/references/preflight-checklist.md`) — Core Web Vitals + WCAG 2.2 AA. Every build. 2. **anti-slop self-checklist** (`atelier-perf-a11y/references/anti-slop-pr