
zhizhuodemao/js-reverse-mcp
Developer toolsThis page was translated automatically. Technical details may differ from the original. View the original
AI Agent-first JS reverse MCP Server: includes header Chrome debugging, breakpoints, network/WebSocket analysis, Patchright reverse detection, optional CloakBrowser.
Overview
AI-first / AI-native JavaScript reverse engineering MCP Server enables your AI coding assistants (such as Claude, Cursor, Copilot) to continuously debug, locate, save, and review JavaScript behavior in web pages like an analyst. It doesn't simply transfer the Chrome DevTools API to the model, but reorganizes scripts, breakpoints, networks, WebSockets, browser states, and local file I/O into tools suitable for AI Agent continuous inference and operations. Anti-detection is part of this capability: by default, it is based on the Patchright protocol layer stealth, and for strong anti-crawling sites, CloakBrowser source code layer fingerprinting mode can be optionally enabled. - : Tool granularity, output boundaries, and error prompts are all designed around Agent decisions to avoid pushing the model to the next invalid step - : Script source code, raw network data, and binary results can all be exported to local files for subsequent analysis input - : When paused, values can be evaluated directly in the call frame, scope variables checked, executed step by step, and returned to source context - : List all loaded JS, search code, retrieve/save source code, automatically format large compressed scripts - : Request call stack, XHR breakpoints, Set-Cookie identification, raw body/header export, WebSocket message grouping - : Clean current site cookies / cache / storage / sessionStorage, coordinated with reload to reproduce cookies and risk control processes - : Browsers visible, cookies / localStorage retain across sessions - :P atchright protocol layer stealth enabled by default; For strong anti-scraping sites, you can add --cloak using CloakBrowser binary Go to Cursor Settings -> MCP -> New MCP Server and use the configuration above. The core goal of this project is not to "operate browsers," but to enable AI Agents to reliably complete a round of real JS reverse engineering: opening pages, passing risk control, locating scripts, saving source code, setting breakpoints, triggering behavior, checking runtime, exporting network materials, reproducing states, and then continuing inference.
Overview (machine translated)
English | Chinese
AI-first / AI-native JavaScript reverse engineering MCP Server enables your AI coding assistants (such as Claude, Cursor, Copilot) to continuously debug, locate, save, and review JavaScript behavior in web pages like an analyst.
It does not simply transfer the Chrome DevTools API to the model as is, but rather reorganizes scripts, breakpoints, networks, WebSockets, browser states, and local file I/O into tools suitable for continuous inference and operations by AI Agents. Anti-detection is part of this capability: by default, Patchright fork is recompiled and published for this MCP, and protocol layer stealth is provided; for strong anti-crawling sites, CloakBrowser source code layer fingerprint mode can be optionally enabled.
☁️ Sponsorship · Sponsored by IPWO
IPWO residential proxies support flexible proxy configurations, allowing developers to select appropriate network environments for different tasks for web access, request testing, localized content analysis, and automated workflows.
For projects involving JS reverse engineering, browser debugging, and web analysis, residential proxies can also serve as a supplementary option to network environments, helping developers flexibly meet access and testing needs in different regions.
👉 Free test entry, 10% off promo code: 0204
☁️ Sponsored · Sponsored by Infistar.cc Infinite Galaxy
**js-reverse-mcp × Infistar.cc Infinite Galaxy | Full Model API · Stable driver with AI-assisted debugging **
Thanks to Infistar.cc Infinite Galaxy for sponsoring and providing model service support to js-reverse-mcp!
- ⚡ Stable handling of complex debugging tasks: Offers enterprise-grade high concurrency channels and multi-node redundancy, with prices as low as 10% of official channels, reducing bandwidth limiting, 429, and long task disconnection issues.
- 🧠 One API Key to Mainstream Models: Supports models such as ChatGPT, Claude, Gemini, Kimi, GLM, DeepSeek, and adapts AI programming tools like Claude Code, Codex, and Cursor.
- 🔎 Supporting Authorization Debugging and Security Research: Suitable for multi-step tasks such as script retrieval, breakpoint analysis, network request tracing, call stack checking, and code logic understanding.
- 📦 Exclusive Benefits for Project Users: Register through the exclusive promotion link and complete your first call to receive a $5 equivalent test quota / exclusive first-top discount.
Positioning: Claude Code / Codex model API configuration support.
Functional Features
- AI-native Tool Design: Tool granularity, output boundaries, and error prompts are all designed around Agent decisions to avoid pushing the model to an invalid next step
- Replayable Workflow: Script source code, raw network data, and binary results can all be exported to local files for subsequent analysis input
This page was translated automatically. Technical details may differ from the original.
Install
npx js-reverse-mcpConfiguration
{
"mcpServers": {
"js-reverse": {
"command": "npx",
"args": ["js-reverse-mcp"]
}
}
}