Open-source CLI that records browser video proof when AI coding agents build UI features, closing the blind-build loop.
An independent developer launched ProofShot, an open-source CLI tool that integrates with AI coding agents (Claude Code, Cursor, Codex, Gemini CLI, Windsurf, GitHub Copilot) to provide visual verification of UI features. The tool records browser sessions, captures screenshots, collects console/server errors, and bundles artifacts for human review. It ships with a headless Chromium browser, installs via npm, and includes a GitHub PR integration command to post evidence inline — with no cloud dependency required.
AI coding agents write code without any feedback loop on visual output — they can't see if the button renders, the form submits, or the console is throwing errors. ProofShot inserts a Playwright-backed browser session between agent output and human review, recording timestamped video with action markers and error logs. It installs at the user level across all projects in two npm commands, with no per-project config needed.
Run `npm install -g proofshot && proofshot install` in your current dev environment, then trigger your agent to build a UI feature and wrap it with `proofshot start --run 'npm run dev' --port 3000` — you get a video + error report before you even open the browser yourself.
Run: npm install -g proofshot && proofshot install
Tags
Sources
Also today
Signals by role
Also today
Tools mentioned