Loading ToolBench
增强版 JS 逆向分析 MCP Server - 提供强大的网站加密算法分析能力
Server exhibits significant definition quality gaps. Tool names lack clear action verbs (analyze_website, smart_analyze are vague). Descriptions are present but generic (avg ~80 chars, below 194-char baseline). Input schemas are visible and properly typed, but output schemas are completely undocumented. No error handling guidance. Parameters lack actionable constraints, e.g., 'focus' enum values are listed but not explained in descriptions. Tool composition is poor: analyze_website, smart_analyze, and analyze_with_browser overlap significantly without clear differentiation. No pagination support despite tools returning potentially large result sets. Security concerns: tools accept arbitrary URLs and execute code (beautify_js, deobfuscate_js, generate_python_code) without input validation or sandboxing guidance.
分析 API 请求的参数和签名
分析网站的 JS 加密算法和签名机制
在真实浏览器环境中访问网站并动态分析 JS 加密算法
美化混淆的 JS 代码
自动识别并绕过验证挑战（五秒盾、Cloudflare 等）
反混淆 JavaScript 代码
检测 JS 代码中使用的加密算法
检测网站使用的反爬虫保护措施
No output schemas documented. LLMs cannot plan downstream tool calls or extract required fields. All 13 tools lack return type documentation.
Tool naming lacks clear action verbs. 'analyze_website', 'smart_analyze', 'analyze_with_browser' are vague and overlapping. LLMs cannot distinguish when to use each. Should be 'analyze_website_static', 'analyze_website_dynamic', 'analyze_website_smart' or similar.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 52 | <=2025-11-25 | v2 |
从网页中提取所有 JS 文件
根据 JS 算法生成对应的 Python 实现代码
生成完整的逆向分析报告
智能分析网站，自动选择最优分析策略
验证签名算法的正确性
Parameter descriptions are generic and lack actionable constraints. E.g., 'focus' enum values (signature, encryption, request_headers, all) are listed but not explained. 'depth' enum (quick, standard, deep) lacks guidance on when to use each. No format specs for 'url' parameter (must be valid HTTP URL?).
No error handling guidance. Tools accept arbitrary URLs and execute code (beautify_js, deobfuscate_js, generate_python_code) without documented error cases or recovery paths. What happens if URL is unreachable? If JS code is malformed? If Python generation fails?
No pagination support. Tools like 'extract_js_files' and 'analyze_website' may return large result sets but lack limit/offset/page parameters. Large responses will exhaust context windows and degrade LLM reasoning.