Loading…
A professional agent skill that audits any website or web app across 160+ checkpoints
No reviews yet — be the first to review this product.
Initial release — 160+ checkpoints across 27 categories, with a launch gate, JSON report schema and fix playbook.
Jun 26, 2026Included in your Pro membership. Cancel anytime.
Agent skill
SKILL.md· no dependencies
Works with
Audits any website or web app across 160+ checkpoints — security, payments integrity, auth, data protection, AI/LLM safety, performance, accessibility, SEO, privacy/legal, reliability and UX — phase by phase, with evidence-backed verdicts, a hard launch gate, and a machine-readable report.
Example prompt
Audit my Next.js + Supabase app before launch — is it launch-ready? Then fix the CRITICAL findings.
Read-only analysis. Reads your project files and/or live site to gather evidence; never fabricates findings, code, URLs or legal text; marks anything it cannot verify as UNVERIFIED.
Get the files
Copy the SKILL.md or download the full bundle (SKILL.md + references) with Pro.
Add it to your skills directory
Put the launch-audit-pro/ folder in your agent's skills path — e.g. ~/.claude/skills/ for Claude Code, or the equivalent for Cursor / Codex.
Run it
Ask your agent: “Run a launch-readiness audit of this project.” The skill triggers automatically.
Bundle (5 files)
---
name: launch-audit-pro
description: Professional, end-to-end launch-readiness audit — 160+ checkpoints
across security, payments, auth, data protection, AI/LLM safety, performance,
accessibility (WCAG 2.2 AA), SEO, privacy & legal, and UX.
---
# Launch Audit Pro
A 160+ checkpoint launch-readiness audit for websites and web apps. It scores
19 engineering/security/compliance categories and 8 UX/product categories,
classifies every finding by severity, applies a hard launch gate, and outputs
both a human report and a machine-readable result.
## Execution Protocol — run phase by phase, verify at every step
1. Evidence or it didn't happen — every PASS/FAIL cites a file:line, config key,
network response or header.
2. Never guess — mark UNVERIFIED and state exactly what would settle it.
3. No fabrication — never invent paths, code, URLs, quotes, stats or legal text.
4. Always know where you stand — keep the progress ledger updated.
5. Audit silently, report once.
### Phases
Phase 0 — Scope & inventory · Phase 1 — Recon & live walkthrough ·
Phase 2 — Category passes (A1→A19, then B1→B8) · Phase 3 — Self-verification ·
Phase 4 — Report · Phase 5 — Fix · Phase 6 — Re-audit**Launch Audit Pro** is a professional, end-to-end launch-readiness audit for websites and web apps — 160+ checkpoints across security, payments integrity, authentication, data protection, AI/LLM safety, reliability, observability, performance & Core Web Vitals, accessibility (WCAG 2.2 AA), SEO, privacy & legal compliance, and UX. Built for founders, indie hackers and agencies shipping products — especially those assembled quickly with AI assistance, where client-side paywalls, exposed databases, leaked secrets, prompt-injection and broken consent banners quietly lose revenue. **What makes it different** - 27 categories (19 engineering/security/compliance · 8 UX/product) and 160+ checkpoints — well beyond the typical ~90-point checklist. - Disciplined phased execution with a visible progress ledger and a mandatory self-verification pass that actively tries to disprove its own CRITICAL findings. - Evidence-or-UNVERIFIED rule eliminates hallucinated findings; unverifiable security controls count as risk, not as a pass. - Severity model + hard launch gate (any open CRITICAL = BLOCKED) for a real go/no-go. - Ships a human report template, a JSON result schema (draft-07) for CI/dashboards, and a fix playbook with secure, server-side remediation patterns. - Honesty rules: never fabricates legal text, testimonials, stats or URLs. Works with Claude Code / Cowork and any SKILL.md-compatible agent. No dependencies — model-driven. **Includes:** SKILL.md, README, and references (report-template.md, report-schema.json, fix-playbook.md).
| Pro | Access while subscribed. | Included in Pro |
|---|
What you get