Ahtasham00/tactab

Overview

Tactile browser control and multimodal visual vision bridge for AI agents (Cursor, Claude, etc.) via Model Context Protocol (MCP) and a Chrome Extension. It is a JavaScript-based open-source project that connects AI agents to a browser for control and visual understanding.

Best for

Best for
Developers prototyping AI browser control with MCP and Chrome.

Use cases

• Let an AI agent control a Chrome browser via MCP
• Give Cursor or Claude visual awareness of web pages
• Build multimodal browser automation workflows

How to use

Tools exposed

• take_screenshot
• scrape_data
• click_button
• fill_form
• get_elements
• get_html

Tested with

Claude Desktop, Cursor, Windsurf, Cline, Continue, ChatGPT

Notes

Tactile browser control and multimodal visual vision bridge for AI agents (Cursor, Claude, etc.) via Model Context Protocol (MCP) and a Chrome Extension. It is a JavaScript-based open-source project that connects AI agents to a browser for control and visual understanding.

0 stars on GitHub. Last updated 2026-10-02. Licensed MIT.

Use cases

• Let an AI agent control a Chrome browser via MCP
• Give Cursor or Claude visual awareness of web pages
• Build multimodal browser automation workflows

Pros

• Uses standard MCP for agent integration
• Combines browser control with visual understanding
• Open source JavaScript codebase

Cons

• Zero GitHub stars suggests limited adoption
• Requires both MCP server and Chrome extension setup
• No evidence of maintenance or documentation beyond description

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.