[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M71KCWX27Q8V8GTETN0PXKTS":3,"skill-same:01M71KCWX27Q8V8GTETN0PXKTS":73},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":23,"publishedAt":44,"linkOnly":45,"markdownContent":46,"markdownBody":47,"frontMatter":48,"refName":49,"commitSha":50,"skillDirectory":51,"forksCount":52,"repositoryHtmlUrl":53,"repositoryHomepageUrl":54,"ownerHtmlUrl":55,"ownerBlogUrl":56,"ownerDescription":57,"updatedAt":58,"install":59,"resources":72},"01M71KCWX27Q8V8GTETN0PXKTS","dsl-vm-reverse",null,"Reverse JavaScript-based custom DSL\u002FVM interpreters, non-standard WASM-like runtimes, and risk-control engines. Use when analyzing IIFE or switch-based opcode dispatchers, extracting instruction tables, recovering bytecode semantics, capturing VM state at runtime, or reconstructing execution flow.","coco-research\u002Fcoco","coco-research","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F68911604?v=4","01M71JRE9C4NM4NSZM4JD0BJTB","Coco Research","01M71JREA0ZMR0PMH3CKA99SN1",631,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Freverse-skill\u002Fskills\u002Freverse-engineering\u002Fdsl-vm-reverse\u002FSKILL.md",[18],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43],"vm","dsl","reverse","dsl vm reverse","bytecode","javascript","when","standard","custom","at","tables","extracting","runtimes","runtime","switch","use","risk","execution","flow","analyzing","2026-11-10 10:12:30",false,"---\nname: dsl-vm-reverse\ndescription: Reverse JavaScript-based custom DSL\u002FVM interpreters, non-standard WASM-like runtimes, and risk-control engines. Use when analyzing IIFE or switch-based opcode dispatchers, extracting instruction tables, recovering bytecode semantics, capturing VM state at runtime, or reconstructing execution flow.\n---\n\n# 🔄 DSL 自定义虚拟机逆向（DSL VM Reverse Engineering）\n\n## 目录\n\n- [1. 适用范围](#1-适用范围)\n- [2. DSL VM 识别特征](#2-dsl-vm-识别特征)\n- [3. 通用逆向工作流](#3-通用逆向工作流)\n- [4. Opcode 提取与分类](#4-opcode-提取与分类)\n- [5. 运行时捕获方案](#5-运行时捕获方案)\n- [6. 常见状态码](#6-常见状态码)\n- [7. Skill 自检清单](#7-skill-自检清单)\n\n---\n\n## 1. 适用范围\n\n当目标文件符合以下 **任意特征** 时使用本 skill：\n\n| # | 特征 | 说明 |\n|---|------|------|\n| 1 | IIFE 开头 + 大量单字母变量名 | `!function(){var U=void 0,y=parseInt,E0=Function,...}` |\n| 2 | 包含 `DG()` 或类似函数含 switch-case 循环 | 解释器主循环，`d[7]&31` 解码 opcode |\n| 3 | 大文件（500KB+）但零字节占比 \u003C 1% | 非标准 WASM，纯 JS |\n| 4 | 包含 `C[number]` 常量表引用 | `C[9][xxx]` 函数表\u002F字符串表 |\n| 5 | 单行压缩代码 | 583KB 单行，混淆变量名 |\n\n### 排除规则\n\n| 条件 | 非本 skill | 转至 |\n|------|-----------|------|\n| 文件以 `\\x00asm` 开头 | 标准 WASM 二进制 | `reverse-engineering\u002Flanguages.md` |\n| 文件以 `Uint8Array([0,97,115,109])` 含 WASM 魔术字 | WASM 嵌入式 | 提取 .wasm 后转 IDA\u002FGhidra |\n| 标准 Webpack 打包（`function(e,t,n){...}`） | 普通 JS | `js-reverse\u002F` |\n| 零字节占比 > 20% | WASM 二进制 | `reverse-engineering\u002Flanguages.md` |\n\n---\n\n## 2. DSL VM 识别特征\n\n### 代码特征\n\n```javascript\n\u002F\u002F 特征 1: IIFE 入口，单字母变量映射数字常量\n!function(){\n    var U=void 0, y=parseInt, E0=Function, AN=Uint8Array;\n    var E=15, l=10, m=12, x=16, S=13, $=11;\n    \u002F\u002F 数字常量映射为变量名，替代原始数字\n    ...\n}\n\n\u002F\u002F 特征 2: 解释器主循环 DG()\nfunction DG(C, d, ...) {\n    var d = [];  \u002F\u002F 数组模拟 WASM stack\u002Flocals\n    for (d[7] = x; d[7] !== U;) {\n        var aE = d[7] & 31;         \u002F\u002F 低 5 位 = opcode\n        var O = d[7] >> 5 & 31;      \u002F\u002F 高 5 位 = sub-operation\n        switch (aE) {\n            case 0: \u002F* ... *\u002F d[7] = 612; break;\n            case 1: \u002F* ... *\u002F\n            \u002F\u002F ... N 个 case\n        }\n    }\n}\n\n\u002F\u002F 特征 3: 常量表 C[9] 存储函数索引和字符串\n\u002F\u002F C[9][0] = [\"pc\"]      → 函数参数描述\n\u002F\u002F C[9][667] = \"string\"  → 字符串常量\n\u002F\u002F C[9][x] = number      → 函数索引\n\n\u002F\u002F 特征 4: W(C[index], null, ...) 调用模式\n\u002F\u002F W = Function.prototype.call.bind(call)\n\u002F\u002F 所有内置函数通过 C[index] 索引调用\n\n\u002F\u002F 特征 5: 指令编码格式\n\u002F\u002F d[7] = opcode(bit 0-4) | subop(bit 5-9) | operand(bit 10+)\n```\n\n### Opcode 编码格式\n\n每条指令编码为 32 位整数：\n\n```\nbit 0-4:   opcode (0-N)\nbit 5-9:   sub-operation (0-31)\nbit 10-31: operand\u002F立即数\n\n解码:\n  aE = d[7] & 31        → opcode\n  O  = d[7] >> 5 & 31   → sub-operation\n  d[other] = d[7] >> 10  → operand\n```\n\n---\n\n## 3. 通用逆向工作流\n\n### Phase 1: 文件分类（5 分钟）\n\n```bash\n# 检查是否为 DSL VM\npython3 \u003C\u003C 'EOF'\nwith open('target.js', 'rb') as f:\n    head = f.read(100)\n\n# 1. 检查 WASM 魔术字\nif head[:4] == b'\\x00asm':\n    print(\"标准 WASM 二进制\")\n    exit()\n\n# 2. 检查零字节占比\ndata = open('target.js', 'rb').read()\nzero_pct = data.count(b'\\x00') \u002F len(data) * 100\nprint(f\"零字节占比: {zero_pct:.1f}%\")\n\nif zero_pct > 20:\n    print(\"WASM 二进制\")\nelif head[:2] == b'!f':\n    # 检查单字母变量模式\n    if b'var U=void 0' in head or b'U=void 0,y=parseInt' in head:\n        print(\"→ DSL VM!\")\n    else:\n        print(\"普通 JS IIFE\")\nEOF\n```\n\n### Phase 2: 变量映射表提取（10 分钟）\n\n```python\nimport re\n\nwith open('target.js', 'r', errors='replace') as f:\n    s = f.read()\n\n# 提取开头 2000 字符的 var X=数字 映射\nmappings = re.findall(r'var\\s+(\\w+)\\s*=\\s*(\\d+)', s[:2000])\nprint('常量映射:')\nfor name, val in mappings:\n    print(f\"  {name:4s} = {val:3d} (0x{int(val):02x})\")\n```\n\n### Phase 3: Opcode 提取与分类（15 分钟）\n\n```python\n# 1. 提取所有 case\nall_cases = re.findall(r'case\\s+(\\d+):', s)\nunique = sorted(set(int(c) for c in all_cases))\n\nprint(f\"总 case: {len(all_cases)} 个\")\nprint(f\"唯一 opcode: {len(unique)} 个: {unique}\")\n\n# 2. 分类每个 opcode\nfor op in unique:\n    idx = s.find(f'case {op}:')\n    snippet = s[idx:idx+200]\n    if 'd[7]=' in snippet:\n        op_type = 'BRANCH'\n    elif 'return' in snippet:\n        op_type = 'RETURN'\n    elif 'W(C[' in snippet:\n        op_type = 'CALL'\n    elif 'new' in snippet:\n        op_type = 'ALLOC'\n    elif 'try' in snippet or 'catch' in snippet:\n        op_type = 'EXCEPTION'\n    else:\n        op_type = 'ARITH\u002FSTORE'\n    print(f\"  opcode {op:2d}: {op_type}\")\n```\n\n### Phase 4: 常量表分析（30 分钟）\n\n```python\nconst_refs = re.findall(r'C\\[9\\]\\[(\\d+)\\]', s)\nunique_refs = sorted(set(int(x) for x in const_refs))\n\nprint(f\"C[9] 引用: {len(unique_refs)} 个索引\")\nprint(f\"范围: {min(unique_refs)} - {max(unique_refs)}\")\n\n# 对每个引用分析上下文\nfor ref in unique_refs[:20]:\n    idx = s.find(f'C[9][{ref}]')\n    ctx = s[max(0,idx-50):idx+80]\n    clean = ''.join(c if c.isprintable() else ' ' for c in ctx)\n    print(f\"  C[9][{ref}] → {clean}\")\n```\n\n### Phase 5: 导出函数追踪（1-2 小时）\n\n导出函数（如 `getToken`）通过以下路径定位：\n\n```\n1. 找 AWSCInner.register() 或类似注册调用\n2. 确定注册的模块和工厂函数\n3. 找工厂函数返回的对象 → 导出函数定义位置\n4. 若函数名不在 JS 中 → 在 C[9] 常量表中作字节码存储\n5. 追踪调用链:\n   AWSCInner._modules['fy'].getToken()\n   → W(C[函数索引], null, ...)\n   → DG() 解释器执行编码后的指令序列\n```\n\n### Phase 6: 运行时注入（若纯静态分析不够）\n\n```javascript\n\u002F\u002F 注入最小 AWSC 兼容环境\nconst fakeEnv = {\n    AWSCInner: {\n        _modules: {},\n        register(name, moduleName, factory) {\n            this._modules[moduleName] = factory();\n        }\n    }\n};\n\n\u002F\u002F 执行 DSL VM 代码\ndslVmCode();\n\n\u002F\u002F 获取导出\nconst token = fakeEnv.AWSCInner._modules['fy'].getToken({});\n```\n\n---\n\n## 4. Opcode 提取与分类\n\n### 参考 opcode 对照表（基于已有案例）\n\n| Opcode | 操作类型 | 特征 |\n|--------|---------|------|\n| 0 | **BRANCH** | `d[7]=xxx` 无条件跳转 |\n| 1 | **CALL** | `W(C[Y],null,function(){...})` 嵌入函数调用 |\n| 2 | **ARITH** | `d[4]=0`, `d[7]=72` 变量赋值 |\n| 3 | **ARITH** | `d[0]=d[1][C[x]]`, `d[5]=d[0]\u003Cd[3]` 比较运算 |\n| 4 | **STORE** | `d[8]=d[5]in d[4]` 属性访问\u002F存在检查 |\n| 5 | **ARITH** | `d[8]=d[4]-d[8]` 算术运算 |\n| 6 | **RETURN** | `return gV`, `throw` 返回\u002F抛出异常 |\n| 7 | **ALLOC** | `d[6]=[]`, `d[6][C[8]](...)` push 操作 |\n| 8 | **BRANCH** | `d[7]=d[k]?512:425` 条件跳转 |\n| 9 | **STRING** | `d[6][C[t]]=d[m]`, `new fh(...)` 正则 |\n| 10 | **ALLOC** | 函数参数准备、调用栈创建 |\n| 11 | **STRING** | `new fh(\"\\\\s\",d[5])` 正则匹配 |\n| 12 | **STORE** | `P[d[9]]=d[4][C[H]](d[3])` 数据传递 |\n| 13 | **CALL** | `C[9][113]=d[9]` 模块初始化 |\n| 14 | **STRING** | `d[8]=d[9]+d[m]` 字符串拼接 |\n| 15 | **RETURN** | `return EL;` 函数返回 |\n| 16 | **ALLOC** | `var r,P,Z,B...` 局部变量声明 |\n| 17 | **ALLOC** | `(Z=[])[C[8]](69,T,445)` 静态数组初始化 |\n| 18 | **TABLE** | 函数表\u002F类型表初始化 |\n| 19 | **EXCEPTION** | `try{for(var RK=x;...` try-catch 循环 |\n| 20 | **DOM** | `Is[d[o]]` DOM 操作 |\n| 21 | **STORE** | 安全获取全局\u002F对象属性 |\n| 22 | **STRING** | `new fh(r,v)` 字符串\u002F正则处理 |\n| 23 | **BRANCH** | `try...catch` 安全获取 + 条件跳转 |\n| 24 | **CALL** | `W(C[2],null,8,z,FL)` 多参数函数调用 |\n| 25 | **EXCEPTION** | `try{...}catch(C){...}` 异常捕获 + 跳转 |\n\n---\n\n## 5. 运行时捕获方案\n\n### 方案 A: Selenium + CDP 原生事件（推荐，成功率最高）\n\n```python\nfrom selenium import webdriver\n\ndriver = webdriver.Chrome()\n\n# 注入反检测\ndriver.execute_cdp_cmd(\"Page.addScriptToEvaluateOnNewDocument\", {\n    \"source\": r\"\"\"\n        Object.defineProperty(navigator, 'webdriver', {get: () => false});\n        Object.defineProperty(navigator, 'plugins', {get: () => [1,2,3,4,5]});\n        Object.defineProperty(navigator, 'languages', {get: () => ['zh-CN','zh','en']});\n    \"\"\"\n})\n\n# 发送 CDP 原生鼠标事件\ndriver.execute_cdp_cmd(\"Input.dispatchMouseEvent\", {\n    \"type\": \"mousePressed\",\n    \"x\": 549.5, \"y\": 441.2,\n    \"button\": \"left\", \"buttons\": 1,\n    \"clickCount\": 1, \"pointerType\": \"mouse\"\n})\n```\n\n### 方案 B: Playwright 无头浏览器\n\n```javascript\nconst { chromium } = require('playwright');\n\nasync function run() {\n    const browser = await chromium.launch();\n    const page = await browser.newPage();\n\n    \u002F\u002F 拦截网络请求\n    await page.route('**\u002Fapi\u002F**', async route => {\n        await route.continue_();\n    });\n\n    await page.goto('https:\u002F\u002Ftarget-page.com');\n\n    \u002F\u002F 等待 DSL VM 初始化\n    await page.waitForFunction(() => {\n        return window.AWSCInner &&\n               window.AWSCInner._modules &&\n               window.AWSCInner._modules['fy'];\n    });\n\n    \u002F\u002F 执行操作\n    await page.mouse.move(500, 400);\n    await page.mouse.down();\n    \u002F\u002F ... 操作序列\n    await page.mouse.up();\n}\n```\n\n### 方案 C: 纯协议验证（成功率极低）\n\n> DSL VM 生成的 token 通常与浏览器上下文强绑定（TLS JA3 指纹、IP、Cookie、请求头等），脱离浏览器后服务端可检测到上下文不匹配。**不建议使用纯协议方案**。\n\n---\n\n## 6. 常见状态码\n\n| Code | 含义 | 处理 |\n|------|------|------|\n| 0 | **验证通过** ✅ | 取出 sessionId + sig |\n| 300 | **风控拦截** | 被拦截，无法通过 |\n| 8778 | **验证失败，需重试** | 重试操作 |\n| 8776 | **操作太快，需重试** | 增加延迟后重试 |\n| 69634 | **通用失败** | 检查参数是否正确 |\n\n---\n\n## 7. Skill 自检清单\n\n- [ ] 我是否完成了 DSL VM 识别（IIFE + 单字母变量 + DG() 解释器）？\n- [ ] 我是否提取了变量映射表（`var X=数字`）？\n- [ ] 我是否提取了 opcode 列表并分类？\n- [ ] 我是否分析了常量表 C[9] 的引用范围？\n- [ ] 我是否定位了导出函数注册点？\n- [ ] 纯静态分析不够时，我是否尝试了运行时注入方案？\n\n---\n","# 🔄 DSL 自定义虚拟机逆向（DSL VM Reverse Engineering）\n\n## 目录\n\n- [1. 适用范围](#1-适用范围)\n- [2. DSL VM 识别特征](#2-dsl-vm-识别特征)\n- [3. 通用逆向工作流](#3-通用逆向工作流)\n- [4. Opcode 提取与分类](#4-opcode-提取与分类)\n- [5. 运行时捕获方案](#5-运行时捕获方案)\n- [6. 常见状态码](#6-常见状态码)\n- [7. Skill 自检清单](#7-skill-自检清单)\n\n---\n\n## 1. 适用范围\n\n当目标文件符合以下 **任意特征** 时使用本 skill：\n\n| # | 特征 | 说明 |\n|---|------|------|\n| 1 | IIFE 开头 + 大量单字母变量名 | `!function(){var U=void 0,y=parseInt,E0=Function,...}` |\n| 2 | 包含 `DG()` 或类似函数含 switch-case 循环 | 解释器主循环，`d[7]&31` 解码 opcode |\n| 3 | 大文件（500KB+）但零字节占比 \u003C 1% | 非标准 WASM，纯 JS |\n| 4 | 包含 `C[number]` 常量表引用 | `C[9][xxx]` 函数表\u002F字符串表 |\n| 5 | 单行压缩代码 | 583KB 单行，混淆变量名 |\n\n### 排除规则\n\n| 条件 | 非本 skill | 转至 |\n|------|-----------|------|\n| 文件以 `\\x00asm` 开头 | 标准 WASM 二进制 | `reverse-engineering\u002Flanguages.md` |\n| 文件以 `Uint8Array([0,97,115,109])` 含 WASM 魔术字 | WASM 嵌入式 | 提取 .wasm 后转 IDA\u002FGhidra |\n| 标准 Webpack 打包（`function(e,t,n){...}`） | 普通 JS | `js-reverse\u002F` |\n| 零字节占比 > 20% | WASM 二进制 | `reverse-engineering\u002Flanguages.md` |\n\n---\n\n## 2. DSL VM 识别特征\n\n### 代码特征\n\n```javascript\n\u002F\u002F 特征 1: IIFE 入口，单字母变量映射数字常量\n!function(){\n    var U=void 0, y=parseInt, E0=Function, AN=Uint8Array;\n    var E=15, l=10, m=12, x=16, S=13, $=11;\n    \u002F\u002F 数字常量映射为变量名，替代原始数字\n    ...\n}\n\n\u002F\u002F 特征 2: 解释器主循环 DG()\nfunction DG(C, d, ...) {\n    var d = [];  \u002F\u002F 数组模拟 WASM stack\u002Flocals\n    for (d[7] = x; d[7] !== U;) {\n        var aE = d[7] & 31;         \u002F\u002F 低 5 位 = opcode\n        var O = d[7] >> 5 & 31;      \u002F\u002F 高 5 位 = sub-operation\n        switch (aE) {\n            case 0: \u002F* ... *\u002F d[7] = 612; break;\n            case 1: \u002F* ... *\u002F\n            \u002F\u002F ... N 个 case\n        }\n    }\n}\n\n\u002F\u002F 特征 3: 常量表 C[9] 存储函数索引和字符串\n\u002F\u002F C[9][0] = [\"pc\"]      → 函数参数描述\n\u002F\u002F C[9][667] = \"string\"  → 字符串常量\n\u002F\u002F C[9][x] = number      → 函数索引\n\n\u002F\u002F 特征 4: W(C[index], null, ...) 调用模式\n\u002F\u002F W = Function.prototype.call.bind(call)\n\u002F\u002F 所有内置函数通过 C[index] 索引调用\n\n\u002F\u002F 特征 5: 指令编码格式\n\u002F\u002F d[7] = opcode(bit 0-4) | subop(bit 5-9) | operand(bit 10+)\n```\n\n### Opcode 编码格式\n\n每条指令编码为 32 位整数：\n\n```\nbit 0-4:   opcode (0-N)\nbit 5-9:   sub-operation (0-31)\nbit 10-31: operand\u002F立即数\n\n解码:\n  aE = d[7] & 31        → opcode\n  O  = d[7] >> 5 & 31   → sub-operation\n  d[other] = d[7] >> 10  → operand\n```\n\n---\n\n## 3. 通用逆向工作流\n\n### Phase 1: 文件分类（5 分钟）\n\n```bash\n# 检查是否为 DSL VM\npython3 \u003C\u003C 'EOF'\nwith open('target.js', 'rb') as f:\n    head = f.read(100)\n\n# 1. 检查 WASM 魔术字\nif head[:4] == b'\\x00asm':\n    print(\"标准 WASM 二进制\")\n    exit()\n\n# 2. 检查零字节占比\ndata = open('target.js', 'rb').read()\nzero_pct = data.count(b'\\x00') \u002F len(data) * 100\nprint(f\"零字节占比: {zero_pct:.1f}%\")\n\nif zero_pct > 20:\n    print(\"WASM 二进制\")\nelif head[:2] == b'!f':\n    # 检查单字母变量模式\n    if b'var U=void 0' in head or b'U=void 0,y=parseInt' in head:\n        print(\"→ DSL VM!\")\n    else:\n        print(\"普通 JS IIFE\")\nEOF\n```\n\n### Phase 2: 变量映射表提取（10 分钟）\n\n```python\nimport re\n\nwith open('target.js', 'r', errors='replace') as f:\n    s = f.read()\n\n# 提取开头 2000 字符的 var X=数字 映射\nmappings = re.findall(r'var\\s+(\\w+)\\s*=\\s*(\\d+)', s[:2000])\nprint('常量映射:')\nfor name, val in mappings:\n    print(f\"  {name:4s} = {val:3d} (0x{int(val):02x})\")\n```\n\n### Phase 3: Opcode 提取与分类（15 分钟）\n\n```python\n# 1. 提取所有 case\nall_cases = re.findall(r'case\\s+(\\d+):', s)\nunique = sorted(set(int(c) for c in all_cases))\n\nprint(f\"总 case: {len(all_cases)} 个\")\nprint(f\"唯一 opcode: {len(unique)} 个: {unique}\")\n\n# 2. 分类每个 opcode\nfor op in unique:\n    idx = s.find(f'case {op}:')\n    snippet = s[idx:idx+200]\n    if 'd[7]=' in snippet:\n        op_type = 'BRANCH'\n    elif 'return' in snippet:\n        op_type = 'RETURN'\n    elif 'W(C[' in snippet:\n        op_type = 'CALL'\n    elif 'new' in snippet:\n        op_type = 'ALLOC'\n    elif 'try' in snippet or 'catch' in snippet:\n        op_type = 'EXCEPTION'\n    else:\n        op_type = 'ARITH\u002FSTORE'\n    print(f\"  opcode {op:2d}: {op_type}\")\n```\n\n### Phase 4: 常量表分析（30 分钟）\n\n```python\nconst_refs = re.findall(r'C\\[9\\]\\[(\\d+)\\]', s)\nunique_refs = sorted(set(int(x) for x in const_refs))\n\nprint(f\"C[9] 引用: {len(unique_refs)} 个索引\")\nprint(f\"范围: {min(unique_refs)} - {max(unique_refs)}\")\n\n# 对每个引用分析上下文\nfor ref in unique_refs[:20]:\n    idx = s.find(f'C[9][{ref}]')\n    ctx = s[max(0,idx-50):idx+80]\n    clean = ''.join(c if c.isprintable() else ' ' for c in ctx)\n    print(f\"  C[9][{ref}] → {clean}\")\n```\n\n### Phase 5: 导出函数追踪（1-2 小时）\n\n导出函数（如 `getToken`）通过以下路径定位：\n\n```\n1. 找 AWSCInner.register() 或类似注册调用\n2. 确定注册的模块和工厂函数\n3. 找工厂函数返回的对象 → 导出函数定义位置\n4. 若函数名不在 JS 中 → 在 C[9] 常量表中作字节码存储\n5. 追踪调用链:\n   AWSCInner._modules['fy'].getToken()\n   → W(C[函数索引], null, ...)\n   → DG() 解释器执行编码后的指令序列\n```\n\n### Phase 6: 运行时注入（若纯静态分析不够）\n\n```javascript\n\u002F\u002F 注入最小 AWSC 兼容环境\nconst fakeEnv = {\n    AWSCInner: {\n        _modules: {},\n        register(name, moduleName, factory) {\n            this._modules[moduleName] = factory();\n        }\n    }\n};\n\n\u002F\u002F 执行 DSL VM 代码\ndslVmCode();\n\n\u002F\u002F 获取导出\nconst token = fakeEnv.AWSCInner._modules['fy'].getToken({});\n```\n\n---\n\n## 4. Opcode 提取与分类\n\n### 参考 opcode 对照表（基于已有案例）\n\n| Opcode | 操作类型 | 特征 |\n|--------|---------|------|\n| 0 | **BRANCH** | `d[7]=xxx` 无条件跳转 |\n| 1 | **CALL** | `W(C[Y],null,function(){...})` 嵌入函数调用 |\n| 2 | **ARITH** | `d[4]=0`, `d[7]=72` 变量赋值 |\n| 3 | **ARITH** | `d[0]=d[1][C[x]]`, `d[5]=d[0]\u003Cd[3]` 比较运算 |\n| 4 | **STORE** | `d[8]=d[5]in d[4]` 属性访问\u002F存在检查 |\n| 5 | **ARITH** | `d[8]=d[4]-d[8]` 算术运算 |\n| 6 | **RETURN** | `return gV`, `throw` 返回\u002F抛出异常 |\n| 7 | **ALLOC** | `d[6]=[]`, `d[6][C[8]](...)` push 操作 |\n| 8 | **BRANCH** | `d[7]=d[k]?512:425` 条件跳转 |\n| 9 | **STRING** | `d[6][C[t]]=d[m]`, `new fh(...)` 正则 |\n| 10 | **ALLOC** | 函数参数准备、调用栈创建 |\n| 11 | **STRING** | `new fh(\"\\\\s\",d[5])` 正则匹配 |\n| 12 | **STORE** | `P[d[9]]=d[4][C[H]](d[3])` 数据传递 |\n| 13 | **CALL** | `C[9][113]=d[9]` 模块初始化 |\n| 14 | **STRING** | `d[8]=d[9]+d[m]` 字符串拼接 |\n| 15 | **RETURN** | `return EL;` 函数返回 |\n| 16 | **ALLOC** | `var r,P,Z,B...` 局部变量声明 |\n| 17 | **ALLOC** | `(Z=[])[C[8]](69,T,445)` 静态数组初始化 |\n| 18 | **TABLE** | 函数表\u002F类型表初始化 |\n| 19 | **EXCEPTION** | `try{for(var RK=x;...` try-catch 循环 |\n| 20 | **DOM** | `Is[d[o]]` DOM 操作 |\n| 21 | **STORE** | 安全获取全局\u002F对象属性 |\n| 22 | **STRING** | `new fh(r,v)` 字符串\u002F正则处理 |\n| 23 | **BRANCH** | `try...catch` 安全获取 + 条件跳转 |\n| 24 | **CALL** | `W(C[2],null,8,z,FL)` 多参数函数调用 |\n| 25 | **EXCEPTION** | `try{...}catch(C){...}` 异常捕获 + 跳转 |\n\n---\n\n## 5. 运行时捕获方案\n\n### 方案 A: Selenium + CDP 原生事件（推荐，成功率最高）\n\n```python\nfrom selenium import webdriver\n\ndriver = webdriver.Chrome()\n\n# 注入反检测\ndriver.execute_cdp_cmd(\"Page.addScriptToEvaluateOnNewDocument\", {\n    \"source\": r\"\"\"\n        Object.defineProperty(navigator, 'webdriver', {get: () => false});\n        Object.defineProperty(navigator, 'plugins', {get: () => [1,2,3,4,5]});\n        Object.defineProperty(navigator, 'languages', {get: () => ['zh-CN','zh','en']});\n    \"\"\"\n})\n\n# 发送 CDP 原生鼠标事件\ndriver.execute_cdp_cmd(\"Input.dispatchMouseEvent\", {\n    \"type\": \"mousePressed\",\n    \"x\": 549.5, \"y\": 441.2,\n    \"button\": \"left\", \"buttons\": 1,\n    \"clickCount\": 1, \"pointerType\": \"mouse\"\n})\n```\n\n### 方案 B: Playwright 无头浏览器\n\n```javascript\nconst { chromium } = require('playwright');\n\nasync function run() {\n    const browser = await chromium.launch();\n    const page = await browser.newPage();\n\n    \u002F\u002F 拦截网络请求\n    await page.route('**\u002Fapi\u002F**', async route => {\n        await route.continue_();\n    });\n\n    await page.goto('https:\u002F\u002Ftarget-page.com');\n\n    \u002F\u002F 等待 DSL VM 初始化\n    await page.waitForFunction(() => {\n        return window.AWSCInner &&\n               window.AWSCInner._modules &&\n               window.AWSCInner._modules['fy'];\n    });\n\n    \u002F\u002F 执行操作\n    await page.mouse.move(500, 400);\n    await page.mouse.down();\n    \u002F\u002F ... 操作序列\n    await page.mouse.up();\n}\n```\n\n### 方案 C: 纯协议验证（成功率极低）\n\n> DSL VM 生成的 token 通常与浏览器上下文强绑定（TLS JA3 指纹、IP、Cookie、请求头等），脱离浏览器后服务端可检测到上下文不匹配。**不建议使用纯协议方案**。\n\n---\n\n## 6. 常见状态码\n\n| Code | 含义 | 处理 |\n|------|------|------|\n| 0 | **验证通过** ✅ | 取出 sessionId + sig |\n| 300 | **风控拦截** | 被拦截，无法通过 |\n| 8778 | **验证失败，需重试** | 重试操作 |\n| 8776 | **操作太快，需重试** | 增加延迟后重试 |\n| 69634 | **通用失败** | 检查参数是否正确 |\n\n---\n\n## 7. Skill 自检清单\n\n- [ ] 我是否完成了 DSL VM 识别（IIFE + 单字母变量 + DG() 解释器）？\n- [ ] 我是否提取了变量映射表（`var X=数字`）？\n- [ ] 我是否提取了 opcode 列表并分类？\n- [ ] 我是否分析了常量表 C[9] 的引用范围？\n- [ ] 我是否定位了导出函数注册点？\n- [ ] 纯静态分析不够时，我是否尝试了运行时注入方案？\n\n---\n",{"name":5,"description":7},"main","cd89f013b617f7eea982a9bbc60e992974fda2aa","systems\u002Freverse-skill\u002Fskills\u002Freverse-engineering\u002Fdsl-vm-reverse",15,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco","https:\u002F\u002Fcocoresearch.org\u002Fcoco\u002F","https:\u002F\u002Fgithub.com\u002Fcoco-research","https:\u002F\u002Fcocoresearch.org","Local-first, open-source tools for people who work with AI agents: governed memory, agents that check their own work, and data that stays on your machine.","2026-11-11 12:18:13",{"source":8,"skillName":5,"commands":60,"prompt":70,"zipAvailable":71},[61,64,67],{"tool":62,"command":63},"npx","npx skills add coco-research\u002Fcoco --skill dsl-vm-reverse",{"tool":65,"command":66},"bunx","bunx skills add coco-research\u002Fcoco --skill dsl-vm-reverse",{"tool":68,"command":69},"pnpm","pnpm dlx skills add coco-research\u002Fcoco --skill dsl-vm-reverse","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Freverse-skill\u002Fskills\u002Freverse-engineering\u002Fdsl-vm-reverse\u002FSKILL.md",true,[],[74,99,109,126,161,180],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":81,"publishedAt":44},"01M71KEJXF9KMT0BWCMRSRFA4M","trading-super-intelligence","Trading Super Intelligence Team. Named real-world personas across 8 cells. Built local-first (LM Studio) and validator-gated. Illustrative composites; see DISCLAIMER.md.","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Ftrading\u002FSKILL.md",[80],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98],"super","trading super intelligence","intelligence","trading","validator","personas","local","composites","studio","team","across","world","first","gated","2026","lm studio","local-first",{"skillNo":100,"name":101,"nameZh":6,"description":102,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":106,"publishedAt":44},"01M71KEFWS3CMAP7DF1P0FWTDE","strategy-super-intelligence","Strategy Super Intelligence Team. Named real-world personas across 7 cells. Built local-first (LM Studio) and validator-gated. Illustrative composites; see DISCLAIMER.md.","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Fstrategy\u002FSKILL.md",[105],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[107,82,108,84,86,87,88,89,90,91,92,93,94,95,96,97,98],"strategy","strategy super intelligence",{"skillNo":110,"name":111,"nameZh":6,"description":112,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":113,"categories":114,"keywords":121,"publishedAt":44},"01M71KED56ZEPK2QE84QSQPMT0","risk-compliance-super-intelligence","Risk & Compliance (GRC) Super Intelligence Team. Named real-world personas across 7 cells. Built local-first (LM Studio) and validator-gated. Illustrative composites; see DISCLAIMER.md.","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Frisk-compliance\u002FSKILL.md",[115,116],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},{"categoryNo":117,"categoryName":118,"taxonomyType":119,"taxonomyTypeDesc":120},"01JBSEEDCAT000000000000037","律师与法务","PROFESSION","适用职业",[40,82,122,123,124,84,86,87,88,89,125,90,91,92,93,94,95,96,97,98],"risk compliance super intelligence","compliance","risk compliance","grc",{"skillNo":127,"name":128,"nameZh":6,"description":129,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":130,"categories":131,"keywords":145,"publishedAt":44},"01M71KE9A2FWSC74VA96W1EZSN","product-design-super-intelligence","Your product + design brain trust and decision partner. Named personas across 8 cells spanning product strategy, discovery\u002Fresearch, growth\u002Fmetrics, design foundations\u002Fusability, design leadership\u002Fcraft, UX\u002Fcontent, design systems, and a…","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Fproduct-design\u002FSKILL.md",[132,135,138,139,142],{"categoryNo":133,"categoryName":134,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000015","视觉设计",{"categoryNo":136,"categoryName":137,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000021","资料检索与研究",{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},{"categoryNo":140,"categoryName":141,"taxonomyType":119,"taxonomyTypeDesc":120},"01JBSEEDCAT000000000000031","产品经理",{"categoryNo":143,"categoryName":144,"taxonomyType":119,"taxonomyTypeDesc":120},"01JBSEEDCAT000000000000032","设计师",[146,147,84,148,82,149,150,151,152,153,154,91,92,155,156,96,157,158,159,160],"design","product","product design","product design super intelligence","foundations","product-design","decision","design systems","usability","leadership","ux","brain","craft","metrics","trust",{"skillNo":162,"name":163,"nameZh":6,"description":164,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":165,"categories":166,"keywords":169,"publishedAt":44},"01M71KE6H9FPSD9TECYFBPXZ8Y","legal-ip-super-intelligence","Your legal, IP, and regulatory brain trust. 31 named personas across 7 cells covering patent strategy, tech licensing, regulatory litigation, AI governance, open-source compliance, privacy law, and antitrust.","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Flegal-ip\u002FSKILL.md",[167,168],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},{"categoryNo":117,"categoryName":118,"taxonomyType":119,"taxonomyTypeDesc":120},[170,82,171,84,172,173,160,174,87,175,176,177,107,91,92,178,123,179,96,157],"ip","legal ip super intelligence","legal","sep","tech","litigation","ai governance","open","open-source","governance",{"skillNo":181,"name":182,"nameZh":6,"description":183,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":184,"categories":185,"keywords":187,"publishedAt":44},"01M71KE36QRY610V6YVB9JWRND","healthcare-life-sciences-super-intelligence","Your healthcare and life sciences brain trust. 25 named personas across 7 cells covering drug discovery, medtech, biotech strategy, health policy, genomics, clinical trials, and healthcare economics.","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Fhealthcare-life-sciences\u002FSKILL.md",[186],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[188,82,189,190,84,173,160,191,192,87,193,194,107,91,92,195,196,197,198,96],"healthcare","healthcare life sciences super intelligence","life","economics","discovery","policy","clinical","clinical trials","health","trials","genomics"]