claw-mcp-toolkit

Instructions

Implementation Reference

• src/modules/web.ts:20-33 (handler)

  The handler for the 'web_extract_links' tool, which fetches a URL's HTML, uses a regex to find all 'href' attributes starting with 'http', de-duplicates them, and returns up to 30 unique links.

  server.tool("web_extract_links", "Extract all links from a webpage", {
    url: z.string().url().describe("URL to extract links from")
  }, async ({ url }) => {
    const html = await safeFetchText(url);
    const links: string[] = [];
    const regex = /href=["']([^"']+)["']/gi;
    let match;
    while ((match = regex.exec(html)) !== null) {
      const href = match[1];
      if (href.startsWith("http")) links.push(href);
    }
    const unique = [...new Set(links)].slice(0, 30);
    return { content: [{ type: "text", text: `**Links from** ${url}\n\n${unique.map((l, i) => `${i + 1}. ${l}`).join("\n")}\n\nTotal: ${unique.length} unique links` }] };
  });