About screencli

screencli is an AI reviewer that tests what shipped, not what's in the diff. Install the GitHub App and every pull request gets an AI-driven browser test on your preview deploy — screencli reads the change, opens the preview in a real browser, clicks through it like a human, and posts a pass/fail verdict with a recording as a PR comment, before your review even starts.

AI coding agents are becoming the primary way developers build software, and they ship faster than anyone can review. Every check passing — the code compiles, the units don't lie, the deploy went up — still tells you nothing about whether a human can actually use the change. screencli fills that gap by verifying the thing that matters: does the feature work in the browser?

The same engine runs from the command line. Point the CLI at any URL from your terminal or an AI coding agent (Claude Code, Cursor, Windsurf, and more) for one-off verifications, demos, and recordings — it drives an AI browser session and outputs a polished, shareable video with auto-trim, zoom, click highlights, gradient backgrounds, and platform-ready exports. The GitHub App is automatic QA on every PR; the CLI is the same capability on demand.

The CLI is free, with all video effects, gradient backgrounds, and export presets and no limits. The cloud at screencli.sh adds the hosted GitHub App, verdict history, shareable links, hosted rendering, and analytics.