claude-eye — MCP server that lets AI coding agents see and interact with a real browser (via Playwright), with a click-to-pick element selector so you can point at exactly what to change instea
An MCP server that gives any AI CLI (Claude Code, Kiro, etc.) the ability to see, interact with, and accurately inspect localhost web apps in a real browser.
AI coding agents normally can't see what's actually rendered in the browser — they guess based on code alone. This MCP server drives a real (headed) Chromium instance via Playwright, so the agent can navigate, click, type, read computed CSS, check accessibility/contrast, and — most importantly — let the user visually pick an exact element by clicking it, instead of the agent guessing which button/input you mean.
- mcp-browser-server/ — the MCP server (TypeScript, Playwright) - test-page/ — a small HTML fixture used for testing the tools
Register it in your MCP client config (e.g. ~/.claude/.mcp.json):
Tool Description ------ browsernavigate Navigate to a URL browserscreenshot Screenshot the page or one element browsergetdom Get fully rendered HTML browsergetaccessibilitytree Get the ARIA tree browserqueryelements Find elements + properties by CSS selector browsergetcomputedstyles Exact computed CSS for one element browsergetcomputedstylesall Exact computed CSS for all matches browsergetconsolelogs Captured console messages / JS errors browsergetnetworkerrors Captured failed requests (4xx/5xx) browserclearlogs Clear captured logs browserclick Click an element browsertype Type into an input/textarea browserhover Hover an element browserscroll Scroll to (x, y) browserwaitfor Wait for a selector to appear browserevaluate Run arbitrary JS in the page browsersetviewport Resize viewport (responsive testing) browsercheckco
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/nevil06--claude-eye.html)
ChatGPT thinks. Codex works. Use ChatGPT as the planning brain while keeping the Codex harness.
shinthink/blitzstrike⚡ Blitz Strike — a universal MCP penetration-testing toolbelt. Structured methodology: reconnaissance & attack-surface mapping, source-to-sink analysis, and live validation. 57 escalation chains, 130-
0xwilliamortiz/openclaude-improvedruns anywhere. uses anything
halofyai/halofyHalofy is the open access and governance layer for AI agents across your organization. Identity, policy, provenance, audit, and signed erasure.
CaptureGrubEnchant/SolidWorksSolidWorks MCP Server connects an AI assistant to a running SolidWorks instance. Sketch, extrude, fillet, export STEP/STL, generate macros.
arielshad/3d-asset-server3d assets api and mcp
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.