Klavis AI is building open-source infrastructure to make Model Context Protocols (MCPs) easy for everyone. We provide:

• 💬 Slack & Discord Clients: Run MCPs directly from your favorite messaging platforms
• ☁️ Hosted MCP Servers: Access powerful tools without infrastructure management
• 🎛️ Simple Web UI: Configure and manage everything with no coding required

Whether you're a non-technical user wanting to leverage AI workflows or a developer looking to build and scale MCPs, Klavis makes it simple.

See how easy it is to use MCPs (like our Report Generator, YouTube tools, and Document Converters) directly from Slack/Discord:

You can find more case study videos on our website.

For detailed MCP client setup instructions, please refer to the platform-specific documentation:

• Discord Bot Setup Guide - Complete instructions for setting up and running the Discord bot
• Slack Bot Setup Guide - Step-by-step guide for creating a Slack app and connecting it

For detailed MCP server setup instructions, please refer to the README file under each server folder.

• Discord: For Discord API integration
• Document Conversion (Pandoc): Convert between various file formats using Pandoc
• Firecrawl: For web crawling and data collection
• Firecrawl Deep Research: For deeper web research tasks
• GitHub: Perform GitHub repository operations
• Markitdown: Markdown conversion tool
• Postgres: For PostgreSQL database operations
• Report Generation: Create professional web reports from user query
• Resend: For email services
• Slack: For Slack API integration
• Supabase: For database operations
• YouTube: Extract and convert YouTube video information

Klavis consists of two main components:

Located in the mcp_servers/ directory, these service-specific modules expose capabilities as tools:

• Report Generation: Create professional reports from various data sources
• YouTube: Download, analyze, and transform YouTube content
• Document Conversion: Convert between various file formats using Pandoc
• GitHub: Perform GitHub repository operations
• Slack: For Slack API integration
• Supabase: For database operations
• Firecrawl: For web crawling and data collection
• Resend: For email services
• Postgres: For PostgreSQL database operations
• Discord: For Discord API integration

Located in the mcp_clients/ directory, these client applications connect to MCP servers and interface with end-user platforms:

• Discord Bot: Interactive AI assistant for Discord
• Slack Bot: Interactive AI assistant for Slack
• Base Client: Shared functionality for all platform clients

1. Create a new directory in mcp_servers/
2. Implement the MCP server interface
3. Register your tools with appropriate schemas
4. Connect to your client through the standard SSE protocol

1. Create a new client module in mcp_clients/
2. Extend the base_bot.py functionality
3. Implement platform-specific message handling
4. Connect to MCP servers using mcp_client.py

We love contributions! Join our Discord community to discuss ideas and get help.

If you use Klavis in your research or project, please cite:

@software{klavis2024,
  author = {Klavis AI},
  title = {Klavis: Open-Source Infrastructure for Model Context Protocols},
  year = {2024},
  publisher = {GitHub},
  url = {https://github.com/klavis-ai/klavis}
}

This project is licensed under the MIT License - see the LICENSE file for details.