Telegram MCP Server

# Changelog All notable changes to this project will be documented in this file. ## [0.3.0] - 2025-06-19 ### Added - **Telegram API Integration**: Direct API access using MTProto protocol - **New API-based tools**: - `telegram_api_login` - Authenticate with Telegram API using your API credentials - `api_scrape_channel` - Fast and reliable channel scraping using API - `api_search_channel` - Search within channels for specific keywords - `api_get_channel_info` - Get detailed channel information - `api_logout` - Disconnect from Telegram API - **API Features**: - Much faster data retrieval (100x faster than web scraping) - Access to all message metadata (views, reactions, forwards, edits) - Search functionality within channels - No browser automation needed - Session persistence - authenticate once, use forever - Access to private channels you're a member of - **Manual Scraping Mode**: New `scrape_manual` tool for browser-based manual navigation - **Setup guide**: Comprehensive API_SETUP.md for getting Telegram API credentials ### Changed - Version bumped to 0.3.0 to reflect major API integration - Package now includes `telegram` library for MTProto protocol - Added `input` library for interactive authentication - Improved data parser to filter out UI elements in authenticated scraping ### Fixed - Web scraping now properly filters out UI elements - Better message detection in authenticated Telegram Web - Improved scrolling mechanism for authenticated views ## [0.2.0] - 2025-06-19 ### Added - **Authentication Support**: Implemented browser automation with cookies for accessing restricted content - New `telegram_login` tool for authenticating with Telegram Web - New `telegram_logout` tool for clearing authentication cookies - New `telegram_auth_status` tool for checking authentication status - New `scrape_channel_authenticated` tool for scraping with full access - **Cookie Management**: Persistent session storage in user's AppData directory - **Enhanced Content Detection**: Improved detection and labeling of restricted posts - **Multiple URL Format Support**: Tries embedded, widget, and preview formats - **Authenticated HTML Parser**: Specialized parser for Telegram Web authenticated sessions ### Changed - **Authentication is now default**: All scraping tools automatically use authenticated session if logged in - Improved content extraction with multiple fallback methods - Enhanced error handling for restricted content - Better logging for debugging scraping issues - Upgraded username extraction with multiple detection methods - Tool descriptions updated to reflect automatic authentication usage ### Fixed - Fixed username showing as "unknown" in saved files - Fixed empty posts in markdown output - Fixed view count parsing for K/M suffixes (1.2K → 1200) - Fixed reaction parsing with proper emoji detection - Fixed scrolling direction (now scrolls UP to load older posts) ## [0.1.0] - 2025-06-18 ### Initial Release - Basic Telegram channel/group scraping without API - Puppeteer-based browser automation - Markdown and JSON output formats - Windows compatibility with Claude Desktop - 5 MCP tools: - `scrape_channel` - Limited post scraping - `scrape_channel_full` - Complete channel history - `scrape_group` - Group scraping - `get_channel_info` - Channel metadata only - `scrape_date_range` - Date-filtered scraping