- Home
- Skills
- Code Quality & Review
- App Readiness Audit: 7-Pass Pre-Launch Review and Pen Test
More screenshots
Works with the AI Agents you already use
App Readiness Audit: 7-Pass Pre-Launch Review and Pen Test
Each pass starts with fresh eyes, fixes what it safely can, commits on your working branch and writes its own report.
$29
App Readiness Audit: 7-Pass Pre-Launch Review and Pen Test
Example session with this skill installed
My Expo + Supabase app is going to TestFlight next week. Run the full readiness audit on this repo. I'm on the staging branch and I own the staging server at staging.example.com.
- Read your context and instructions
- Compiled the app readiness audit
The agent confirms the repo is clean and on staging, and asks you to confirm you own the staging host. It copies readiness/ into the repo, commits it, and starts ./readiness/run_readiness.sh.
- The runner prints a time estimate, then runs the seven passes one after another, each in a fresh session.
- Each pass commits its fixes on
staging, for example input validation, missing loading and error states, or a hardened header config. It also tags before and after checkpoints and writes its report intoreadiness/. - Between passes the runner checks history, branches, deleted files, committed
.envfiles and the removed-line count. - If pass 4 needs to know which region your backups live in, the run stops with exit code 2. The question is in
readiness/QUESTIONS_FOR_HUMAN.md, and you resume withSTART=5after answering.
At the end you get readiness/FINAL_SUMMARY.md, one or two phone screens long, with:
- a status line
- one bullet per pass
- the worst issues found and fixed
- a "what you need to do" list labelled hard stop / only you can / needs a professional (for example: publish the privacy policy, turn on database backups, add the App Privacy answers in App Store Connect)
- the exact command to undo any pass
Nothing was pushed, deployed or submitted.
Connects securely to your tools. The creator never sees your data.
About this skill
Seven focused audits, one plain-English to-do list. Point your agent at an app repository and it works through seven passes in order: engineering review, security hardening, design, production readiness, legal, store and web compliance, and an authorized white-box pen test of your own staging environment. Each pass starts with fresh eyes, fixes what it safely can, commits on your working branch and writes its own report. A final session turns everything into readiness/FINAL_SUMMARY.md: what was fixed, and the short list of things only you can do.
Setup is three steps. Work on a practice branch (staging by default), copy the readiness/ folder into your repo and commit it, then run ./readiness/run_readiness.sh with Claude Code, or paste one line into Cursor or another editor agent to run RUN_ALL.md. Any other CLI agent works through AGENT_CMD.
Built for unattended runs.
- The runner starts every pass in a fresh agent session.
- With Claude Code it denies history-changing, branch-switching, push,
ghand common deploy and publish commands. - After every pass it checks in code that history only moved forward, no branch was created or switched, no file outside
readiness/was deleted, no.envfile was committed, and no more than 300 lines were removed outsidereadiness/(configurable). Any violation stops the run with exit code 4 and prints the exact command to go back. It never undoes anything itself. - Every pass tags
checkpoint/before-*andcheckpoint/after-*, so you can roll back one pass and keep the rest.
What each pass covers
| Pass | Focus | Report |
|---|---|---|
| 1 | Correctness, data integrity, reliability, performance, tests, refactor | REVIEW_REPORT.md |
| 2 | Security checklist plus hardening | SECURITY_REPORT.md |
| 3 | Design system, empty/error/loading states, microcopy, accessibility | DESIGN_REPORT.md |
| 4 | Config, backups, monitoring, performance, TLS, SEO, build | PROD_READINESS.md |
| 5 | Privacy policy, terms, consent, data rights, deletion, licenses | LEGAL_READINESS.md |
| 6 | Apple App Store, Google Play and open-web compliance, re-read from the official policy pages each run | STORE_READINESS.md |
| 7 | Authorized white-box pen test of your own staging | PENTEST_REPORT.md |
Honest by design. The passes are told to check current official sources (Apple, Google Play, OWASP, NVD, CISA KEV) rather than rely on memory. They write questions down instead of guessing. They never push, deploy, publish, submit to a store, spend money, rotate credentials or touch DNS or billing; those become NEEDS HUMAN items. If a pass hits a question, the run stops with exit code 2. You write your answer in readiness/QUESTIONS_FOR_HUMAN.md and resume with START=<n>.
Includes an offline test suite for the runner (stub agent, temporary repos, no network): 14 scenarios plus 5 follow-up checks, 19 in all.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Code Quality & Review.
- Results depend on the agent and model you run it with. The prompts forbid guessing and require evidence, but review the reports before acting on them.
- Pass 5 drafts legal documents; it is not legal advice. Have a lawyer review anything you publish.
- Pass 7 is an automated, agent-driven test, not a replacement for a professional penetration test where one is contractually required. Run it only against systems you own or are authorized to test.
- A full run on a medium-sized app can take hours and uses a corresponding amount of your agent's model quota.
- Claude Code CLI flags change between releases; check the flags in
run_readiness.shagainst your installed version. - The deny list for Claude Code uses command prefixes, which unusual spellings of a command can slip past; the post-pass guard in code is the backstop.
- The kit is designed to live in the repository: setup copies
readiness/into the project, and the runner and passes read and write files there.
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 buying a skill.
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 12 days ago
- Software
Git 2.28+Bash 3.2+
The hands-off runner uses the Claude Code CLI or any CLI agent through AGENT_CMD; IDE agents use RUN_ALL.md.
Permissions
File Scopes
The runner script itself makes no network calls unless CHECK_REMOTE=1, which runs git ls-remote origin before and after each pass. It never pushes, deploys, publishes, submits to a store, deletes files or tests, rewrites history, rotates credentials, or changes DNS, billing or third-party accounts. Terminal/Shell: runs run_readiness.sh, git, and the project's own build/test commands. Read Files: reads the whole repository, including config and .env files when auditing. Write Files: commits fixes, writes reports to readiness/ and may write dev/staging database dumps to backups/ (added to .gitignore). Network Access: passes read official docs (Apple, Google Play, OWASP, NVD, CISA KEV, GitHub advisories); pass 7 sends test requests to in-scope hosts the user owns, preferring staging. Environment Variables: the runner reads its settings from AGENT, AGENT_CMD, READINESS_BRANCH, START, MAX_REMOVED_LINES, CHECK_REMOTE.