What is canyon?

This platform collects JavaScript code coverage during end-to-end and UI automation test runs. It separates hit data from map data at compile time, which lets it process the large amounts of coverage data that UI tests generate. The system has three parts: plugins that add instrumentation during CI builds, a Node.js and React application that receives and processes the coverage data, and a web interface for viewing reports broken down by individual test cases. It works with popular build tools like Vite, Next.js, and Webpack, automation frameworks such as Playwright, and CI systems including GitHub Actions and GitLab CI. You can run the community edition yourself on your own hardware using Node.js 20 or later with PostgreSQL 14 or later, or by pulling the zhangtao25/canyon Docker image. The codebase is MIT licensed and receives regular updates.

Should you self-host canyon?

canyon scores 47/100 on selfhost.directory (how scoring works) and ranks #7 of 10 in Testing. It has 178 GitHub stars.

It shipped 12 stable releases in the past year. The latest, v3.0.15, came out on 9 September 2026. New versions land roughly every 3 days.

Setup is advanced: plan for several services and some ongoing maintenance. An official Docker image is available. It is MIT-licensed and written mostly in Typescript.

Its closest alternative on selfhost.directory is Locust (78/100, 28 245 stars). It scores higher overall, so compare the two before you pick. See canyon vs Locust side by side.

Features

Automation framework integrations
Build tool integrations
CI provider detection
GitLab source metadata integration
Per-test-case coverage
Real-time coverage reports
Self-hosted community edition
Separate hit and map data
Source Map Support

FAQ

为什么 Compare 中的变更代码文件数量和 GitLab 不一致？

Compare counts only files that Babel compiles and includes in coverage calculation (js, jsx, ts, tsx), while GitLab shows all Git diff files including docs, images, and config files. The difference is due to different statistical scopes.