Getting started

breakhound finds your framework and dev server (and starts it if needed), discovers every route, sweeps each page from 320px to 1920px in a real browser, finds the exact pixel where things go wrong, and tells you why.

Quick start

In your project folder:

npx @seybel/breakhound

That's it: no config. npx runs it without installing anything. To type breakhound directly instead, install it globally:

npm i -g @seybel/breakhound

Or add it to your project with npm i -D @seybel/breakhound, so everyone gets the same version, and run it as npx breakhound or from an npm script.

If your dev server is already running it's used, otherwise breakhound starts it with your dev script and stops it when done. Chromium is downloaded automatically the first time (about 150 MB).

Other ways to run it:

npx @seybel/breakhound /pricing /blog          # only these routes
npx @seybel/breakhound http://localhost:3000   # just this one page
npx @seybel/breakhound --url https://staging.example.com   # every route, on another server

Finding routes

Project	Where routes come from
Next.js (App and Pages Router)	app/ and pages/, skipping route groups, parallel slots, private folders and API routes
Nuxt	the live Vue router, so module and custom routes count too (falls back to pages/)
SvelteKit	src/routes
Astro	src/pages
Remix, React Router 7	the route manifest, app/routes.ts and app/routes/
TanStack Router	routeTree.gen.ts and the live router
React Router (code-defined)	the live route tree, including <Routes> nested inside pages; falls back to reading your router code
Vue Router (code-defined)	the live router; falls back to reading your router code
Angular	your route definitions, from src/main.ts on, with lazily loaded loadChildren routes under their path
Anything else	links followed from the home page (up to 50 pages, a few per section)

Dynamic routes like /blog/[slug] are scanned once, using a real link found on your pages (routes that share a template share a layout). Routes below one reuse its values: once /users/jane is found, /users/[username]/settings is scanned as /users/jane/settings. If something is still missing, it's listed with a hint naming the parameters. Give them a real value once, by name, and every route that uses them is filled in:

params: { username: 'jane', orderId: '1001' }

Or give one route its own URL, which wins over params:

samples: { '/products/[id]': '/products/42' }

Pages behind a login are spotted (they redirect to a sign-in page) and listed instead of scanning the login page over and over. Sign in once:

npx @seybel/breakhound login

A browser window opens; log in as you normally would, then press Enter. The session is saved to .breakhound/auth.json, which is never committed (see Reports), and used by later scans. breakhound never sees your password.

Links that would log you out, delete something, or download a file are never followed.