- Home
- Skills
- Code Quality & Review
- Audari ship gate
Works with the AI Agents you already use
Audari ship gate
Evidence-based pre-launch audit with a deterministic SHIP / FIX / HOLD verdict for sites and web apps.
$7
Audari ship gate
Example session with this skill installed
We are ready for the MVP launch of our Next.js app at https://acme-notes.example. Can we ship? Analyze the repo and URL in dev mode.
- Read your context and instructions
- Compiled the audari ship gate
- Generated the UI component
I ran Audari ship gate on this repo and https://acme-notes.example (owned). Shape site+app, stage soft-launch / mvp, mode:dev. Nothing critical failed, but required soft-launch items are still open — verdict is
FIX.
Audari ship gate — acme-notes · site+app · soft-launch / mvp
Verdict: FIX · 2026-10-21 · mode: dev
Target: repo ./ @ a1b2c3d · URL https://acme-notes.example (owned: yes) · stage: given
Counts: K 5/5 · G 6/12 · S 5/8 · N/A 1
Do these three first
S8 Abuse control on login (G · FAIL) — done-when: login, signup and reset actions are rate-limited server-side (per IP + per account) · verify: cite the limiter call in app/(auth)/login/actions.ts, then 6 wrong passwords on a staging test account → throttled · owner: you
2.
C5 Social image exists (G · FAIL) — done-when: og:image URL returns 200 image/png · verify: curl.exe -sI https://acme-notes.example/og.png → 200 · owner: you
3.
C6 robots.txt and sitemap.xml (G · FAIL) — done-when: both return 200; robots names the sitemap; sitemap lists only public pages · verify: curl.exe -s https://acme-notes.example/robots.txt · owner: you
Blockers
C7 Custom 404 (G · FAIL) — evidence: /ship-gate-check-404 → 404 with the host's default page; no app/not-found.tsx (only app/r/[token]/not-found.tsx) · done-when: root not-found page with link home · owner: you
S7 Session cookies flagged (G · GAP) — evidence: none; cookie options not set in lib/supabase/server.ts, no Set-Cookie observed · done-when: auth cookies HttpOnly + Secure + SameSite=Lax seen in a staging login response · owner: you
L2 Deployed commit known (G · GAP) — evidence: tree clean at a1b2c3d; production SHA not shown · done-when: Vercel production deployment shows a1b2c3d and every .env.example name exists in Production · owner: you
Should
S10 Baseline security headers (S · FAIL) — evidence: curl.exe -sI shows HSTS and nosniff, no X-Frame-Options and no CSP frame-ancestors · done-when: frame protection header present
- L4 Rollback path known (S · GAP) — done-when: previous Ready production deployment named
- C9 Alt text and focus (S · GAP) — alt present on all 6 home images; focus not tab-tested
N/A
- S3 Admin routes — no admin, debug, internal, seed or cron routes (
rg -loverapp/empty)
OK
- S1 — only
.env.exampletracked; tree and history pattern scan: no matches
S2 — public vars are NEXT_PUBLIC_SUPABASE_URL, NEXT_PUBLIC_SUPABASE_ANON_KEY (publishable); 14 shipped chunks scanned, no matches
S4 — RLS auth.uid() = owner_id on notes (supabase/migrations/0003_rls.sql:12); app/api/notes/[id]/route.ts:9 uses the user-scoped client
S5 — app/api/billing/webhook/route.ts:18 constructEvent with env secret; price IDs from lib/plans.ts:4; access granted in webhook, not on success URL
- L1 — apex and www: valid certificate; no
http://assets on home
S6 — app/error.tsx generic copy; lib/api/handle.ts:7 returns { error: "Something went wrong" }; /api/ship-gate-check → generic 404 JSON
- S9 — preview URL → 401 (deployment protection)
- C1 — contact form: empty / bad email / valid tested in browser; user confirmed delivery (screenshot)
- C2 — 390×844 screenshots of
/and/pricing: no horizontal scroll, CTA visible - C3 — unique titles and descriptions on
/,/pricing,/about - C4 —
/privacy,/terms→ 200; linked inapp/(marketing)/layout.tsx:41 - L3 —
http://→ 308 https;www→ 308 apex - S11 —
package-lock.jsontracked;npm audit --omit=dev: 0 critical, 0 high, 2 moderate - C8 — 11 nav/footer/hero/legal links → 200
- C10 — hero WebP 180 KB; PageSpeed mobile LCP 2.9 s, CLS 0.02
- C11 — user confirmed pageviews in Vercel Analytics
Handoff
- Consent timing for analytics — owner: counsel
- Privacy page content and processor list (Supabase, Stripe, Vercel) — owner: counsel
- Credential-stuffing depth beyond a basic limiter — owner: you
- Backups and a restore drill before paying users — owner: you
Next steps
- Add server-side rate limits on login / signup / reset, then re-check S8.
- Host a real
og.pngthat returns 200 and wire it asog:image. - Ship
robots.txt+sitemap.xmlthat list only public pages. - Re-run Audari ship gate on the same stage; aim for SHIP before ads or indexing push.
Reviewed against the Audari ship gate soft-launch / mvp checks on 2026-10-21.
A dated record of evidence seen — not a certification, legal review, or security audit.
og.png
TSX · React component
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Stop guessing if your web application is ready for production. This skill runs a deterministic pre-launch audit by cross-referencing your repository against your live production URL and returns a clear SHIP, FIX, or HOLD verdict. It moves beyond vague checklists by requiring concrete evidence for every pass, so security basics, site essentials, and deployment health are verified before you widen your user base.
What it does
- Evidence-based verification validates site essentials like robots.txt, sitemap, 404 handling, and social preview images.
- Lean security auditing scans for exposed secrets, open admin/debug routes, object access controls, and payment trust signals.
- Deployment smoke testing checks HTTPS, canonical host, and whether the production deploy/commit can be identified.
- Deterministic reporting generates dev-focused or client-ready launch reports based on a fixed rule set.
- Actionable prioritization identifies the top three items to fix first, ordered by critical risk and blast radius.
How it works
- Scope definition analyzes the repository to set condition flags for features like payments, auth, forms, or server-side logic.
- Evidence collection runs read-only checks against the repo and the production URL (headers and HTML).
- Status assignment marks every gate item as PASS, FAIL, GAP, or N/A based on observed data.
- Verdict application — HOLD if any critical check fails; FIX if required items fail or lack evidence; SHIP if required checks for the stage pass with evidence.
Frameworks & tools
Works with any web stack including React, Next.js, Vue, Svelte, Node.js, and static sites. It requires repo access and an authorized production URL for full verification.
Why this beats prompting it yourself
Generic prompts often hallucinate security or ignore missing evidence. This skill uses a rigid tiering system (Critical, Gate, Should) and fixed verdict logic so the same evidence always results in the same launch decision.
Use cases
- Pre-launch sign-off for developers needing a final check before a soft launch or MVP release.
- Client delivery to provide a professional, plain-language readiness report for non-technical stakeholders.
- Safety check after a major migration to ensure no secrets were exposed and SEO essentials remain intact.
Known limitations
Does not perform penetration testing, load testing, or legal compliance reviews. It cannot fix the identified issues and only supports read-only auditing. Not a GDPR, WCAG, or security certification.
Read-only only. No automated fixing. Does not cover accessibility (WCAG), deep penetration testing, or legal/GDPR compliance. Requires ownership/authorization for URL checks.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 24 days ago
Requires
Permissions
Read-only pre-launch gate: reads the project repo, runs local git/curl checks, and fetches the buyer’s owned production URL. Does not modify files, does not need connected SaaS accounts, and does not read environment secret values to print them.