What is launchworthy?

launchworthy is a Claude Code skill built for people who used an AI tool like Lovable, Bolt, v0, Cursor, or Claude Code itself to build an app, and now want to know if it is actually safe to put in front of real users. Instead of just chatting with an AI and asking it to review your code, this tool runs a structured audit across five specific areas and produces a scored report along with a prioritized list of fixes, each pointing to exact file paths with code you can copy and paste. The five areas it checks are frontend and user experience, backend and data handling, authentication and security, infrastructure and deployment, and operations and recovery. Concretely, this covers things like broken loading states, a database with row level security accidentally turned off, secret keys exposed in the browser bundle, missing backups, and whether you could actually recover if something went wrong. The tool automatically figures out what framework and backend you are using, among options like Next.js, SvelteKit, Astro, Supabase, and Firebase, and adjusts its checks to match. What makes this different from just asking an AI to review your code is that it refuses to mark something as safe unless it can actually verify it. Anything it cannot check itself is labeled as needing manual verification rather than being assumed fine, and if you insist something is okay without proof, it records that as unverified rather than turning the item green. It also pushes back on common excuses like calling something "just an MVP," since those are exactly the situations that tend to lead to real security problems. Getting started involves installing Claude Code, then adding this project as a plugin with two simple commands, after which you can simply ask in plain English whether your app is production ready. It then runs the audit and writes a scored report you can act on. The project is released under the MIT license and works across many different app stacks rather than being tied to one specific framework.