Launchpad

Check an app is safe to launch, before deploying it or whenever asked. Finds the security holes and deploy-breakers AI-built apps ship with (database tables anyone can change, keys in the browser or the repository, AI routes anyone can run up, pages that 404 on the host, imports that fail on Linux) and says how to fix each one. Works on any app folder: no launchpad.json, account or credentials needed. Reads the files git would commit, on this machine; nothing is uploaded.

Create launchpad.json for an app (database and bot protection on) and keep Launchpad's secrets folder out of git.

Show what launchpad_up would create or update, and anything missing first (provider credentials, secrets). Calls no provider.

Create the app's services, wire their keys together and deploy it. Safe to run again: it resumes and updates instead of duplicating. Can take several minutes while a new database starts.

List the resources Launchpad manages for the app.

Probe the deployed app's health URLs and report status codes and response times. (To check the code before launch, use launchpad_launch_check.)

Check which provider credentials the app needs and which are set.

Names of the env vars an environment of the app gets (never their values), and which are custom secrets.