Klavis AI provides open-source, easy-to-use "MCPs" (Modular Control Programs) that integrate AI workflows into Slack, Discord, and web interfaces. It targets both non-technical users seeking to leverage AI and developers aiming to build and scale AI applications, simplifying access to tools like report generation, web crawling, and document conversion.

How It Works

Klavis is architected with two core components: MCP Servers and MCP Clients. MCP Servers, located in mcp_servers/, expose specific capabilities as tools (e.g., YouTube processing, document conversion via Pandoc, web crawling with Firecrawl). These servers communicate with clients via the Server-Sent Events (SSE) protocol. MCP Clients, found in mcp-clients/, are platform-specific applications (Discord Bot, Slack Bot) that connect to MCP servers, enabling end-users to interact with AI tools through familiar messaging platforms or a simple web UI.

Quick Start & Requirements

• Discord Bot Setup: Refer to the Discord Bot Setup Guide.
• Slack Bot Setup: Refer to the Slack Bot Setup Guide.
• MCP Server Setup: Instructions are within each server's folder README (e.g., mcp_servers/discord/README.md).
• Dependencies: Python, Discord API integration, Slack API integration, Pandoc, Firecrawl, PostgreSQL, Resend, Supabase. Specific versions are not detailed in the README.

Highlighted Details

• Supports integration with popular platforms like Slack and Discord.
• Offers a simple web UI for configuration and management.
• Includes a variety of pre-built MCPs for tasks like report generation, YouTube analysis, and document conversion.
• Designed for extensibility, allowing users to add new tools or client platforms.

Maintenance & Community

• The project encourages community contributions and provides a Discord community for discussion.
• A citation is provided in BibTeX format for academic use.

Licensing & Compatibility

• Licensed under the MIT License. This license is permissive and generally compatible with commercial and closed-source applications.

Limitations & Caveats

The README does not specify hardware requirements, detailed dependency versions, or provide performance benchmarks. Setup time and resource footprint are not estimated.