ProofShot – Give AI coding agents eyes to verify the UI they build
FreeFreeFree tier
About ProofShot – Give AI coding agents eyes to verify the UI they build
ProofShot is an open-source CLI tool that gives AI coding agents the ability to visually verify the UI they build. It wraps a development server, records the agent's browser interactions as video, captures console errors, and generates a standalone proof bundle with a synchronized timeline, screenshots, and error logs. The output is designed to be PR-ready, including a SUMMARY.md and visual diffs against baselines. Works with any AI coding agent (Claude Code, Cursor, Codex, Gemini CLI, Windsurf, etc.) and can be installed via npm in seconds.
Key Features
Full session video recording with automatic dead-time trimming
Browser console error detection and server log pattern matching for multiple languages
Interactive standalone HTML viewer with synchronized video, screenshots, and action timeline
PR-ready artifacts: SUMMARY.md, visual diff comparison against baselines
Agent-agnostic – works with Claude Code, Cursor, Codex, OpenCode, Gemini CLI, Windsurf, and any MCP-compatible agent
Three-command workflow: start (launches dev server and headless Chromium), test (agent navigates and clicks), stop (collects errors and generates proof)
Pros & Cons
Pros
- Open source under MIT license – free to use and modify
- Works with any AI coding agent, no vendor lock-in
- Easy to set up with a single npm install command
- Generates comprehensive proof artifacts including video, screenshots, logs, and timeline
- Includes automatic dead-time trimming to focus on actual actions
- Output is PR-ready, reducing manual review effort
Cons
- Requires headless Chromium, which may add system dependencies
- Focused on web UI verification – not suitable for non-browser or CLI-only agent tasks
- As an open-source tool, may lack dedicated enterprise support
Best For
Providing visual proof that an AI coding agent correctly implemented UI changesDebugging AI agent behavior by reviewing recorded sessions and error logsAutomating verification of front-end changes in pull requestsIntegrating visual verification into AI agent development workflows
FAQ
How does ProofShot work?
ProofShot works with three commands: start (launches a dev server and headless Chromium), test (runs agent actions like navigate, click, type, screenshot), and stop (collects errors, stops recording, and generates a proof bundle with video, screenshots, and logs).
Which AI coding agents are supported?
ProofShot is agent-agnostic and works with any AI coding agent, including Claude Code, Cursor, Codex, OpenCode, Gemini CLI, Windsurf, and any MCP-compatible agent.
Is ProofShot free and open source?
Yes, ProofShot is open source under the MIT license and is free to use.
What does the output look like?
The output is a standalone HTML viewer with a synchronized video timeline, screenshots, and action labels. It also includes a SUMMARY.md file and formatted output ready for pull requests, plus visual diff comparisons against baselines.