TikTok MCP Server

Instructions

Implementation Reference

• tiktok_mcp/browser.py:395-450 (handler)

  The actual implementation of the tool handler `get_user_videos` which scrapes TikTok user videos using Playwright.

  async def get_user_videos(self, username: str, count: int = 10) -> list[dict]:
      """Get videos from a specific user's profile."""
      if not username.startswith("@"):
          username = f"@{username}"
  
      page = await self.goto_tiktok(f"/{username}")
      await asyncio.sleep(3)
  
      # Get user info
      user_info = await page.evaluate("""() => {
          const name = document.querySelector('[data-e2e="user-title"]') || document.querySelector('h1');
          const bio = document.querySelector('[data-e2e="user-bio"]');
          const following = document.querySelector('[data-e2e="following-count"]');
          const followers = document.querySelector('[data-e2e="followers-count"]');
          const likes = document.querySelector('[data-e2e="likes-count"]');
          return {
              username: name?.textContent?.trim() || '',
              bio: bio?.textContent?.trim() || '',
              following: following?.textContent?.trim() || '0',
              followers: followers?.textContent?.trim() || '0',
              total_likes: likes?.textContent?.trim() || '0',
          };
      }""")
  
      videos = []
      seen_ids = set()
      scroll_attempts = 0
  
      while len(videos) < count and scroll_attempts < count * 2:
          items = await page.evaluate("""() => {
              const videos = [];
              const cards = document.querySelectorAll('[data-e2e="user-post-item"], [class*="DivItemContainerForSearch"]');
              cards.forEach(card => {
                  try {
                      const link = card.querySelector('a[href*="/video/"]');
                      const views = card.querySelector('[class*="video-count"]') || card.querySelector('strong');
                      const desc = card.querySelector('[title]') || link;
                      
                      const href = link?.href || '';
                      const videoIdMatch = href.match(/video\\/([0-9]+)/);
                      
                      videos.push({
                          video_id: videoIdMatch ? videoIdMatch[1] : null,
                          url: href,
                          description: desc?.title || desc?.textContent?.trim() || '',
                          views: views?.textContent?.trim() || '',
                      });
                  } catch(e) {}
              });
              return videos;
          }""")
  
          for item in items:
              vid = item.get("video_id")
              if vid and vid not in seen_ids:
                  seen_ids.add(vid)

• tiktok_mcp/server.py:236-241 (handler)

  The tool invocation logic for `tiktok_user_videos` inside the `call_tool` function.

  elif name == "tiktok_user_videos":
      results = await browser.get_user_videos(
          arguments["username"],
          arguments.get("count", 10),
      )
      return [TextContent(type="text", text=json.dumps(results, indent=2, ensure_ascii=False))]

• tiktok_mcp/server.py:85-95 (registration)

  The registration of `tiktok_user_videos` tool definition.

  Tool(
      name="tiktok_user_videos",
      description="Get videos from a specific TikTok user's profile.",
      inputSchema={
          "type": "object",
          "properties": {
              "username": {"type": "string", "description": "TikTok username (with or without @)"},
              "count": {"type": "integer", "description": "Number of videos (default 10)", "default": 10},
          },
          "required": ["username"],
      },