nobody writes the tests

which of these you need

Almost none of them, and that is the point. The setup does not vary by stack, because there is nothing to install: you give it a URL that is already running and a sentence describing what should work. npx smolanalytics test --url https://yourapp.com --test "the pricing page shows a monthly price" is the whole thing, and it runs the same against a staging box, a deploy preview, or localhost through a tunnel. No GitHub App, no preview environment for us to build, nothing written to your repo.

What differs by stack is what tends to break, and where the tracking calls live. That is what these pages are for. If you built the app with a coding agent or a builder, you are the sharpest case: fast-moving code nobody read line by line, and no suite to catch a regression. If you are on Next.js, React or Vue, your host already builds a preview URL for every pull request, so the setup is a URL you already have.

The second half is instrumentation, and it uses the same walk. An agent that has just worked out how to check out knows which user actions your app has, so it writes and maintains your tracking calls inside the SDK you already use — PostHog, Mixpanel, Amplitude, Google Analytics, Plausible or Segment. We do not replace your analytics; we keep its instrumentation correct. how that works. If you have no analytics at all, a project can hold its own tracking, as an option rather than the product.

You can skip all of it for now. npx smolanalytics audit reads the repo you are standing in and names the user actions nothing is measuring, with the file and the line, and it needs no account and makes no network call. Neither does the readability check, which answers whether an AI crawler can read a page.

Still deciding? how it works walks the whole loop, the setup docs cover whatever stack is not listed above, the MCP server is how your editor asks its own questions, and pricing is one plan and three numbers.