OSINT MCP Server

Implementation Reference

• src/tools/social-scavenger.ts:13-40 (handler)

  The actual implementation of the Twitter user search logic using a Nitter RSS feed.

  async getTwitterProfile(username: string): Promise<any> {
    try {
      // Using a reliable public nitter instance
      const response = await fetch(`https://nitter.net/${username}/rss`);
      if (!response.ok) throw new Error("Profile not found or instance down");
      const text = await response.text();
  
      // Basic regex extraction from RSS for bio/details
      const titleMatch = text.match(/<title>(.*?)<\/title>/i);
      const descMatch = text.match(/<description>(.*?)<\/description>/i);
  
      return {
        platform: "Twitter/X",
        username,
        title: titleMatch ? titleMatch[1] : "",
        bio: descMatch ? descMatch[1] : "No bio found or private account",
        url: `https://twitter.com/${username}`,
        nitterUrl: `https://nitter.net/${username}`
      };
    } catch (error) {
      return { platform: "Twitter/X", username, error: true, message: (error as Error).message };
    }
  }
  
  // Basic Instagram check (limited due to IG's strict blocking)
  async getInstagramProfile(username: string): Promise<any> {
    try {
      const url = `https://www.instagram.com/${username}/`;

• src/index.ts:732-740 (registration)

  Registration of the twitter_user_search tool in the MCP server.

  server.tool(
    "twitter_user_search",
    { username: z.string().describe("Twitter/X username to lookup") },
    async ({ username }) => {
      const result = await scavengerClient.getTwitterProfile(username);
      return {
        content: [{ type: "text", text: JSON.stringify(result, null, 2) }],
      };
    }