Argos

A passing test suite tells you the code still runs. It says nothing about the button that moved twenty pixels. Argos closes that gap by collecting screenshots from a test run, uploading them as a build and comparing each one against the baseline you approved.

Differences surface as a check on the pull request, and a purpose-built review interface lets a reviewer approve or reject individual changes and comment on specific ones, keeping the GitHub or GitLab check in step with the decision. An approved change becomes the next baseline.

Integrations follow the tools teams already run.

• Playwright: a first-class SDK that captures screenshots from existing tests.
• Storybook: stories are captured and the whole Storybook is hosted on every pull request.
• Other runners: Cypress, Puppeteer and WebdriverIO work against the platform.
• CLI: a command line tool for screenshots produced outside a supported runner.
• MCP server: an agent can review the visual changes its own work produced.
• Debugging: builds keep screenshots, DOM snapshots and Playwright traces, and flaky tests are flagged.

Packages are published on npm per integration and the service works with any CI system. Standalone screenshots and recordings can be shared as links outside a build. The hosted platform has a free tier of five thousand screenshots a month with per screenshot pricing above it, and the repository takes contributions on a branch and pull request basis.