Weibo MCP Server

Instructions

Implementation Reference

• src/server.ts:51-63 (registration)

  Registers the MCP 'get_feeds' tool with input schema for uid and limit, description, and a handler that calls WeiboCrawler.extractWeiboFeeds and returns JSON stringified feeds.

  server.tool("get_feeds",
    "获取指定微博用户的最新动态和帖子",
    { 
      uid: z.number().describe("微博用户的唯一标识符"),
      limit: z.number().describe("返回的最大动态数量")
    },
    async ({ uid, limit }) => {
      const feeds = await crawler.extractWeiboFeeds(uid, limit);
      return {
        content: [{ type: "text", text: JSON.stringify(feeds) }]
      };
    }
  );

• src/weibo.ts:35-60 (handler)

  Core handler implementing the tool logic: fetches user's Weibo feeds by obtaining containerId from profile, then iteratively fetches pages of feeds until the limit is reached or no more available.

  async extractWeiboFeeds(uid: number, limit: number): Promise<Record<string, any>[]> {
    const feeds: Record<string, any>[] = [];
    let sinceId = '';
    
    try {
      const containerId = await this.getContainerId(uid);
      if (!containerId) return feeds;
  
      while (feeds.length < limit) {
        const pagedFeeds = await this.extractFeeds(uid, containerId, sinceId);
        if (!pagedFeeds.Feeds || pagedFeeds.Feeds.length === 0) {
          break;
        }
  
        feeds.push(...pagedFeeds.Feeds);
        sinceId = pagedFeeds.SinceId as string;
        if (!sinceId) {
          break;
        }
      }
    } catch (error) {
      console.error(`无法获取UID为'${uid}'的动态`, error);
    }
    
    return feeds.slice(0, limit);
  }

• src/weibo.ts:118-137 (helper)

  Helper method to extract the containerId for a user's feeds from their profile API response.

  private async getContainerId(uid: number): Promise<string | null> {
    try {
      const response = await axios.get(PROFILE_URL.replace('{userId}', uid.toString()), {
        headers: DEFAULT_HEADERS
      });
      
      const data = response.data;
      const tabsInfo = data?.data?.tabsInfo?.tabs || [];
      
      for (const tab of tabsInfo) {
        if (tab.tabKey === 'weibo') {
          return tab.containerid;
        }
      }
      return null;
    } catch (error) {
      console.error(`无法获取UID为'${uid}'的containerId`, error);
      return null;
    }
  }

• src/weibo.ts:147-169 (helper)

  Helper method to fetch a single page of user feeds using the feeds API URL constructed with containerId and sinceId, returning PagedFeeds structure.

  private async extractFeeds(uid: number, containerId: string, sinceId: string): Promise<PagedFeeds> {
    try {
      const url = FEEDS_URL
        .replace('{userId}', uid.toString())
        .replace('{containerId}', containerId)
        .replace('{sinceId}', sinceId);
      
      const response = await axios.get(url, { headers: DEFAULT_HEADERS });
      const data = response.data;
      
      const newSinceId = data?.data?.cardlistInfo?.since_id || '';
      const cards = data?.data?.cards || [];
      
      if (cards.length > 0) {
        return { SinceId: newSinceId, Feeds: cards };
      } else {
        return { SinceId: newSinceId, Feeds: [] };
      }
    } catch (error) {
      console.error(`无法获取UID为'${uid}'的动态`, error);
      return { SinceId: null, Feeds: [] };
    }
  }

• src/server.ts:53-56 (schema)

  Zod input schema defining parameters for the get_feeds tool: uid (number) and limit (number).

  { 
    uid: z.number().describe("微博用户的唯一标识符"),
    limit: z.number().describe("返回的最大动态数量")
  },