tampermonkey-scripts

Monorepo for userscripts. Each site gets its own folder with clean TypeScript source. On every push to main, GitHub Actions lints, typechecks, and builds — then commits the compiled .user.js files back so they're always available via raw.githubusercontent.com.

Structure

scripts/
  <site-name>/
    src/
      meta.ts        ← @userscript header block (comment only)
      index.ts       ← entry point
      utils.ts       ← helpers, split as needed
    dist/
      <site-name>.user.js   ← compiled output (committed by CI)
    tampermonkey-loader.user.js  ← install this in Tampermonkey once

Adding a new script

1. Copy the example-site folder:

   cp -r scripts/example-site scripts/my-new-site

2. Update meta.ts with the correct @match URLs and script name.
3. Write your code in src/. Split into as many files as you like — esbuild bundles it all.
4. Update tampermonkey-loader.user.js with the new raw URL.
5. Install the loader in Tampermonkey. Done.

Local development

npm install
npm run build          # build all scripts once
npm run build:watch    # rebuild on save
npm run lint           # ESLint
npm run typecheck      # tsc --noEmit

The dist/ files are committed locally too, so you can point the loader at your local dev server or just let CI handle it.

How the loader works

The tampermonkey-loader.user.js file is the only thing you install in Tampermonkey. It fetches the compiled dist/*.user.js from this repo's raw GitHub URL and eval()s it at runtime. This means:

• You never have to touch Tampermonkey again after the initial install
• Iterate freely in src/, commit, CI builds it, the loader picks it up automatically
• Cache-busting via ?_=<timestamp> prevents stale CDN responses during active dev

CI / GitHub Actions

The workflow (.github/workflows/build.yml):

1. Runs on push to main when src/ files change
2. Lints and typechecks
3. Builds all scripts
4. Commits any changed dist/ files back with [skip ci] to prevent loops