@ai-jshook/mcp 0.1.1 → 0.1.3

package/server.json

@@ -1,39 +1,39 @@
-{
-  "$schema": "https://static.modelcontextprotocol.io/schemas/2025-07-09/server.schema.json",
-  "name": "io.github.wuji1/jshook-reverse-tool",
-  "description": "AI-powered JavaScript reverse engineering tool with 80+ MCP tools for browser automation, debugging, network monitoring, and intelligent hooking",
-  "version": "0.1.0",
-  "homepage": "https://github.com/wuji1/jshook-reverse-tool",
-  "license": "MIT",
-  "keywords": [
-    "mcp",
-    "javascript",
-    "reverse-engineering",
-    "browser-automation",
-    "debugging",
-    "deobfuscation",
-    "crypto-detection",
-    "hook",
-    "ai-powered",
-    "puppeteer",
-    "chrome-devtools",
-    "network-monitoring",
-    "anti-detection",
-    "captcha"
-  ],
-  "packages": [
-    {
-      "registry_type": "npm",
-      "identifier": "jshook-reverse-tool",
-      "version": "0.1.0"
-    }
-  ],
-  "author": {
-    "name": "wuji1",
-    "url": "https://github.com/wuji1"
-  },
-  "repository": {
-    "type": "git",
-    "url": "https://github.com/wuji1/jshook-reverse-tool.git"
-  }
-}
+{
+  "$schema": "https://static.modelcontextprotocol.io/schemas/2025-07-09/server.schema.json",
+  "name": "io.github.wuji1/jshook-reverse-tool",
+  "description": "AI-powered JavaScript reverse engineering tool with 80+ MCP tools for browser automation, debugging, network monitoring, and intelligent hooking",
+  "version": "0.1.0",
+  "homepage": "https://github.com/wuji1/jshook-reverse-tool",
+  "license": "MIT",
+  "keywords": [
+    "mcp",
+    "javascript",
+    "reverse-engineering",
+    "browser-automation",
+    "debugging",
+    "deobfuscation",
+    "crypto-detection",
+    "hook",
+    "ai-powered",
+    "puppeteer",
+    "chrome-devtools",
+    "network-monitoring",
+    "anti-detection",
+    "captcha"
+  ],
+  "packages": [
+    {
+      "registry_type": "npm",
+      "identifier": "jshook-reverse-tool",
+      "version": "0.1.0"
+    }
+  ],
+  "author": {
+    "name": "wuji1",
+    "url": "https://github.com/wuji1"
+  },
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/wuji1/jshook-reverse-tool.git"
+  }
+}

package/tsconfig.dev.json

@@ -1,14 +1,14 @@
-{
-  "extends": "./tsconfig.json",
-  "compilerOptions": {
-    "module": "CommonJS"
-  },
-  "ts-node": {
-    "esm": false,
-    "transpileOnly": true,
-    "compilerOptions": {
-      "module": "CommonJS"
-    }
-  }
-}
-
+{
+  "extends": "./tsconfig.json",
+  "compilerOptions": {
+    "module": "CommonJS"
+  },
+  "ts-node": {
+    "esm": false,
+    "transpileOnly": true,
+    "compilerOptions": {
+      "module": "CommonJS"
+    }
+  }
+}
+

package/CLAUDE.md

@@ -1,170 +0,0 @@
-# CLAUDE.md
-
-This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
-
-## Project Overview
-
-**JSHook** 是一个 AI 驱动的 JavaScript 逆向工程工具，提供 100+ 专业 MCP 工具用于浏览器自动化、调试、网络监控和智能 Hook。
-
-### Architecture
-
-```
-jshook-reverse-tool/          # 主项目 (MCP 服务器源码)
-├── src/                       # TypeScript 源码
-├── dist/                      # 编译输出 (MCP 服务器)
-└── CLAUDE.md                 # 本文件
-
-jshook-plugin/                # Claude Code Plugin
-├── .mcp.json                 # MCP 服务器配置
-├── skills/                   # Agent Skills (6个)
-│   ├── browser-automation/
-│   ├── debugger-tools/
-│   ├── network-monitoring/
-│   ├── ai-hooking/
-│   ├── performance-analysis/
-│   └── cache-management/
-└── README.md                 # 插件文档
-```
-
-### Dual Mode Operation
-
-此项目支持两种运行模式：
-
-1. **Standalone Mode** (独立 MCP 服务器)
-   ```bash
-   npm run dev
-   # 启动独立 MCP 服务器，通过 OPENAI_API_KEY/ANTHROPIC_API_KEY 调用 LLM
-   ```
-
-2. **Plugin Mode** (Claude Code Plugin) - **Recommended**
-   ```json
-   // .mcp.json
-   {
-     "mcpServers": {
-       "jshook": {
-         "command": "node",
-         "args": ["../dist/index.js"]
-       }
-     }
-   }
-   ```
-   - MCP 服务器运行在后台
-   - Claude Code 的 Agent Skills 直接调用 MCP 工具
-   - AI 分析由 Claude Code 内置模型处理（无需额外 API Key）
-
-### Plugin Installation
-
-```bash
-# 1. 在 Claude Code 中添加插件
-# 指向 jshook-plugin/.mcp.json
-
-# 2. 使用 Skills 进行逆向工作
-/js-reverse https://example.com/api "header" "X-Signature"
-# 或手动调用 MCP 工具
-browser_launch()
-page_navigate(url="https://example.com")
-```
-
-## Development Commands
-
-```bash
-# 开发模式（自动重启）
-npm run dev
-
-# 构建项目
-npm run build
-
-# 运行测试
-npm test
-
-# 代码检查
-npm run lint
-
-# 代码格式化
-npm run format
-
-# 完整检查
-npm run check
-```
-
-## Architecture
-
-### Source Directory (src/)
-
-```
-src/
-├── index.ts              # MCP 服务器入口
-├── server/               # MCP 服务器层
-│   ├── MCPServer.ts      # 主服务器类 (使用 McpServer API)
-│   ├── BrowserToolHandlers.ts    # 35 浏览器工具
-│   ├── DebuggerToolHandlers.ts   # 38 调试器工具
-│   ├── AIHookToolHandlers.ts     # AI Hook 工具
-│   └── AdvancedToolHandlers.ts   # 网络/性能工具
-├── modules/              # 核心功能模块
-│   ├── collector/        # 代码收集器 (DOM/XHR/脚本)
-│   ├── debugger/         # 调试器 (断点/调用栈)
-│   ├── deobfuscator/     # 反混淆器 (基于规则)
-│   ├── crypto/           # 加密检测 (规则 + AI)
-│   ├── captcha/          # 验证码检测
-│   ├── stealth/          # 反检测脚本
-│   ├── hook/             # Hook 管理器
-│   └── monitor/          # 控制台/性能监控
-├── services/
-│   └── LLMService.ts     # LLM API 封装 (独立模式用)
-├── types/
-│   └── index.ts          # 核心类型定义
-└── utils/
-    ├── TokenBudgetManager.ts  # Token 预算管理
-    ├── UnifiedCacheManager.ts # 统一缓存管理
-    └── logger.ts         # 日志系统
-```
-
-### Plugin Skills (jshook-plugin/skills/)
-
-| Skill | Tools | Description |
-|-------|-------|-------------|
-| `browser-automation/` | 35+ | 页面导航、DOM操作、反检测、CAPTCHA处理 |
-| `debugger-tools/` | 38+ | 断点、调用栈、监控表达式 |
-| `network-monitoring/` | 6+ | HTTP 请求/响应捕获 |
-| `ai-hooking/` | 7 | AI 驱动 Hook 生成 |
-| `performance-analysis/` | 4 | Web Vitals、代码覆盖率 |
-| `cache-management/` | 6 | Token 预算、缓存清理 |
-
-### Key Patterns
-
-1. **MCP Tool Pattern**: 使用 `McpServer.registerTool()` 注册工具
-2. **Zod Schema**: 使用 zod 定义输入输出模式
-3. **Singleton Pattern**: `TokenBudgetManager`、`UnifiedCacheManager`
-4. **ES Modules**: 导入需带 `.js` 扩展名
-5. **Claude Code Integration**: Plugin 模式下，AI 分析由 Claude Code Agent Skills 完成
-
-### Module Dependencies
-
-```
-MCPServer
-├── BrowserToolHandlers → collector + stealth + captcha
-├── DebuggerToolHandlers → DebuggerManager
-├── AIHookToolHandlers → HookManager
-├── Deobfuscator → 规则引擎 (独立模式: LLMService)
-├── CryptoDetector → 规则引擎 (独立模式: LLMService)
-├── TokenBudgetManager (singleton)
-└── UnifiedCacheManager (singleton)
-```
-
-## Tech Stack
-
-| Category | Technology |
-|----------|------------|
-| Runtime | Node.js 18+, TypeScript 5.0+ |
-| Browser Automation | Puppeteer, Chrome DevTools Protocol |
-| MCP Protocol | @modelcontextprotocol/sdk |
-| AST Processing | @babel/parser, @babel/traverse, @babel/types |
-| Testing | Jest, ts-jest |
-
-## Important Notes
-
-1. **Plugin Mode 优先**: 推荐使用 Claude Code Plugin 模式，AI 分析更自然
-2. **独立模式**: 需要 `OPENAI_API_KEY` 或 `ANTHROPIC_API_KEY`
-3. **首次运行**: Puppeteer 会自动下载 Chromium
-4. **工具数量**: 100+ MCP 工具，分布在多个 Handlers 文件
-5. **大数据处理**: 大型数据通过 `detailId` 返回，防止上下文溢出