Weibo MCP Server

Instructions

Implementation Reference

• src/weibo.ts:237-334 (handler)

  The core handler method in WeiboCrawler that implements searching Weibo content: constructs API URL with keyword and page, fetches data, parses cards for type 9 mblogs, extracts text, user info, pics, video, handles pagination up to limit.

  async searchWeiboContent(keyword: string, limit: number, page: number = 1): Promise<ContentSearchResult[]> {
    try {
      const results: ContentSearchResult[] = [];
      let currentPage = page;
      
      while (results.length < limit) {
        const url = SEARCH_CONTENT_URL
          .replace('{keyword}', encodeURIComponent(keyword))
          .replace('{page}', currentPage.toString());
          
        const response = await axios.get(url, {
          headers: DEFAULT_HEADERS
        });
        
        const data = response.data;
        const cards = data?.data?.cards || [];
        
        // 微博通常会返回多个卡片，我们寻找包含微博内容的卡片组
        let contentCards: any[] = [];
        for (const card of cards) {
          // 微博内容卡片通常有card_type=9
          if (card.card_type === 9) {
            contentCards.push(card);
          } 
          // 处理卡片组
          else if (card.card_group && Array.isArray(card.card_group)) {
            const contentGroup = card.card_group.filter((item: any) => item.card_type === 9);
            contentCards = contentCards.concat(contentGroup);
          }
        }
        
        if (contentCards.length === 0) {
          break; // 没有更多内容，退出循环
        }
        
        // 处理每个内容卡片
        for (const card of contentCards) {
          if (results.length >= limit) {
            break;
          }
          
          const mblog = card.mblog;
          if (!mblog) continue;
          
          // 提取图片链接
          const pics: string[] = [];
          if (mblog.pics && Array.isArray(mblog.pics)) {
            for (const pic of mblog.pics) {
              if (pic.url) {
                pics.push(pic.url);
              }
            }
          }
          
          // 提取视频链接
          let videoUrl = undefined;
          if (mblog.page_info && mblog.page_info.type === 'video') {
            videoUrl = mblog.page_info.media_info?.stream_url || 
                       mblog.page_info.urls?.mp4_720p_mp4 ||
                       mblog.page_info.urls?.mp4_hd_mp4 ||
                       mblog.page_info.urls?.mp4_ld_mp4;
          }
          
          // 创建内容搜索结果对象
          const contentResult: ContentSearchResult = {
            id: mblog.id,
            text: mblog.text,
            created_at: mblog.created_at,
            reposts_count: mblog.reposts_count,
            comments_count: mblog.comments_count,
            attitudes_count: mblog.attitudes_count,
            user: {
              id: mblog.user.id,
              screen_name: mblog.user.screen_name,
              profile_image_url: mblog.user.profile_image_url,
              verified: mblog.user.verified
            },
            pics: pics.length > 0 ? pics : undefined,
            video_url: videoUrl
          };
          
          results.push(contentResult);
        }
        
        currentPage++;
        
        // 检查是否有下一页
        if (!data?.data?.cardlistInfo?.page || data.data.cardlistInfo.page === "1") {
          break;
        }
      }
      
      return results.slice(0, limit);
    } catch (error) {
      console.error(`无法搜索关键词为'${keyword}'的微博内容`, error);
      return [];
    }
  }

• src/server.ts:84-97 (registration)

  Registers the MCP tool 'search_content' with name, description, Zod input schema (keyword, limit, optional page), and async handler delegating to WeiboCrawler's searchWeiboContent method, returning JSON stringified results.

  server.tool("search_content",
    "根据关键词搜索微博内容并返回相关的微博帖子",
    { 
      keyword: z.string().describe("搜索微博内容的关键词"),
      limit: z.number().describe("返回的最大微博条目数量"),
      page: z.number().optional().describe("起始页码，默认为1")
    },
    async ({ keyword, limit, page }) => {
      const contents = await crawler.searchWeiboContent(keyword, limit, page || 1);
      return {
        content: [{ type: "text", text: JSON.stringify(contents) }]
      };
    }
  );

• src/types.ts:74-124 (schema)

  TypeScript interface defining the output structure ContentSearchResult for search results, including id, text, timestamps, counts, user details, optional pics and video_url.

  export interface ContentSearchResult {
    /**
     * 微博ID
     */
    id: string;
    
    /**
     * 微博文本内容
     */
    text: string;
    
    /**
     * 创建时间
     */
    created_at: string;
    
    /**
     * 转发数
     */
    reposts_count: number;
    
    /**
     * 评论数
     */
    comments_count: number;
    
    /**
     * 点赞数
     */
    attitudes_count: number;
    
    /**
     * 发布该微博的用户信息
     */
    user: {
      id: number;
      screen_name: string;
      profile_image_url: string;
      verified: boolean;
    };
    
    /**
     * 图片链接列表（如果有）
     */
    pics?: string[];
    
    /**
     * 视频链接（如果有）
     */
    video_url?: string;
  } 

• src/consts.ts:30-30 (helper)

  Constant URL template for Weibo content search API endpoint, with placeholders for {keyword} and {page}, used in searchWeiboContent handler.

  export const SEARCH_CONTENT_URL = 'https://m.weibo.cn/api/container/getIndex?containerid=100103type%3D1%26q%3D{keyword}&page_type=searchall&page={page}';