Agent Skills › NyxFoundation/speca › spec-discovery

spec-discovery

GitHub

从给定URL出发，递归抓取并发现网站中的技术规格文档、白皮书或架构文档链接。通过优先使用静态获取工具，失败时回退至浏览器工具，最终去重输出规范文档列表。

.claude/skills/spec-discovery/SKILL.md NyxFoundation/speca

触发场景

需要查找某项目的技术规格书搜索特定网站的白皮书或协议文档

安装

npx skills add NyxFoundation/speca --skill spec-discovery -g -y

更多选项

非标准路径

npx skills add https://github.com/NyxFoundation/speca/tree/main/.claude/skills/spec-discovery -g -y

不安装直接使用

npx skills use NyxFoundation/speca@spec-discovery

指定 Agent (Claude Code)

npx skills add NyxFoundation/speca --skill spec-discovery -a claude-code -g -y

安装 repo 全部 skill

npx skills add NyxFoundation/speca --all -g -y

预览 repo 内 skill

npx skills add NyxFoundation/speca --list

SKILL.md

Frontmatter

{
    "name": "spec-discovery",
    "context": "fork",
    "description": "Crawl and discover specification documents from a given URL.",
    "allowed-tools": "mcp__fetch__fetch, browser_navigate, browser_scroll, browser_click, browser_view, write"
}

SKILL: Specification Discovery

Mindset

You are a meticulous Web Researcher tasked with finding all relevant technical specification documents starting from a seed URL. Your goal is to be comprehensive and follow all promising links.

Goal

Given a starting URL, navigate the website to find and list all URLs pointing to technical specifications, whitepapers, or architectural documents. These are often found in sections like "Developers", "Documentation", "Technology", or "Whitepaper".

Tools

Primary: mcp__fetch__fetch - Use for static documentation pages (returns Markdown, fast and efficient)
Fallback: Browser tools (browser_navigate, browser_scroll, browser_click, browser_view) - Use for dynamic/JavaScript-rendered pages or when mcp__fetch__fetch fails (403, timeout, JS-required)

Input

A JSON object containing the starting URL:

{
  "url": "https://example.com/project"
}

Procedure

Initial Fetch: Use mcp__fetch__fetch with the provided url to retrieve the page content as Markdown.
Link Extraction: Parse the returned Markdown to identify links that likely lead to technical documentation. Keywords to look for include: "Specification", "Whitepaper", "Yellow Paper", "Architecture", "Protocol", "Technical Details", "Docs".
Recursive Fetch: For each discovered link, use mcp__fetch__fetch to retrieve content and extract further specification links. Limit depth to 2-3 levels.
Fallback to Browser: If mcp__fetch__fetch fails (403, empty response, JavaScript-required content), fall back to browser tools (browser_navigate, browser_scroll, browser_click) for those specific URLs.
Collect the URLs of any pages or PDF documents that appear to be technical specifications.
Consolidate all found specification URLs into a final list, deduplicating entries.

Output Format

Return a JSON object containing a list of found specification URLs. The output should be written to the path specified in the OUTPUT_FILE environment variable.

{
  "start_url": "https://example.com/project",
  "found_specs": [
    {
      "url": "https://example.com/project/docs/specification.md",
      "title": "Project Specification"
    },
    {
      "url": "https://example.com/project/whitepaper.pdf",
      "title": "Project Whitepaper"
    }
  ],
  "metadata": {
    "timestamp": "...",
    "urls_visited": []
  }
}