Telegram MCP Server

Instructions

Implementation Reference

• src/server.ts:478-530 (handler)

  Main handler function that executes the scrape_channel_full tool logic: determines scraper based on auth, scrapes all posts (maxPosts:0), formats sample, reports file save locations.

    private async handleScrapeChannelFull(args: any): Promise<any> {
      // Check if authenticated and use authenticated scraper by default
      const isAuthenticated = await this.auth.isAuthenticated();
      const scraperToUse = isAuthenticated ? this.authScraper : this.scraper;
      
      if (isAuthenticated) {
        logger.info('Using authenticated scraper for full channel scrape (logged in)');
      } else {
        logger.info('Using unauthenticated scraper for full channel scrape (not logged in)');
      }
  
      const options: ScrapeOptions = {
        url: args.url,
        maxPosts: 0, // No limit - get ALL posts
        includeReactions: true
      };
  
      const result = await scraperToUse.scrape(options);
      
      // The scraper already saves to file, so we just need to inform about it
      const channelName = result.channel.username;
      const timestamp = new Date().toISOString().replace(/[:.]/g, '-').slice(0, -5);
      const windowsPath = `C:\\Users\\User\\AppData\\Roaming\\Claude\\telegram_scraped_data\\${channelName}_${timestamp}_full.md`;
      
      // Also return a sample of the content for immediate analysis
      const samplePosts = result.posts.slice(0, 5); // First 5 posts as sample
      const sampleResult = { ...result, posts: samplePosts };
      const sampleMarkdown = this.formatter.format(sampleResult);
      
      return {
        content: [
          {
            type: 'text',
            text: `Successfully scraped ${result.totalPosts} posts from @${channelName}
  
  Files saved to:
  - Markdown: ${windowsPath}
  - JSON: ${windowsPath.replace('.md', '.json')}
  
  Total posts: ${result.totalPosts}
  Date range: ${result.posts.length > 0 ? `${result.posts[result.posts.length - 1]?.date.toISOString().split('T')[0]} to ${result.posts[0]?.date.toISOString().split('T')[0]}` : 'N/A'}
  
  The full channel history has been saved. Here's a sample of the first 5 posts:
  
  ${sampleMarkdown}
  
  To analyze all ${result.totalPosts} posts, open the saved markdown file and copy its contents to Claude.
  
  ${isAuthenticated ? '✅ Scraped using authenticated session - all content including restricted posts should be accessible.' : '⚠️ Scraped without authentication - some restricted content may not be accessible.'}`
          }
        ]
      };
    }

• src/server.ts:148-165 (registration)

  Tool registration in getTools() method, defining name, description, and input schema for scrape_channel_full.

    name: 'scrape_channel_full',
    description: 'Scrape ALL posts from a Telegram channel and save to file. Uses authenticated session if logged in. Returns file location.',
    inputSchema: {
      type: 'object',
      properties: {
        url: {
          type: 'string',
          description: 'The Telegram channel URL (e.g., https://t.me/channelname)'
        },
        save_to_file: {
          type: 'boolean',
          description: 'Save results to MD and JSON files',
          default: true
        }
      },
      required: ['url']
    }
  },

• src/server.ts:150-164 (schema)

  Input schema definition for the scrape_channel_full tool, specifying parameters like url and save_to_file.

  inputSchema: {
    type: 'object',
    properties: {
      url: {
        type: 'string',
        description: 'The Telegram channel URL (e.g., https://t.me/channelname)'
      },
      save_to_file: {
        type: 'boolean',
        description: 'Save results to MD and JSON files',
        default: true
      }
    },
    required: ['url']
  }

• src/server.ts:68-69 (handler)

  Dispatcher case in the CallToolRequestSchema handler that routes to the specific tool handler.

  case 'scrape_channel_full':
    return await this.handleScrapeChannelFull(args);