MCP/llm-fusion-mcp
1
0
Fork 1
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
llm-fusion-mcp/INTEGRATION.md
Ryan Malloy c335ba0e1e
Some checks are pending
🚀 LLM Fusion MCP - CI/CD Pipeline / 🔍 Code Quality & Testing (3.10) (push) Waiting to run
Details
🚀 LLM Fusion MCP - CI/CD Pipeline / 🔍 Code Quality & Testing (3.11) (push) Waiting to run
Details
🚀 LLM Fusion MCP - CI/CD Pipeline / 🔍 Code Quality & Testing (3.12) (push) Waiting to run
Details
🚀 LLM Fusion MCP - CI/CD Pipeline / 🛡️ Security Scanning (push) Blocked by required conditions
Details
🚀 LLM Fusion MCP - CI/CD Pipeline / 🐳 Docker Build & Push (push) Blocked by required conditions
Details
🚀 LLM Fusion MCP - CI/CD Pipeline / 🎉 Create Release (push) Blocked by required conditions
Details
🚀 LLM Fusion MCP - CI/CD Pipeline / 📢 Deployment Notification (push) Blocked by required conditions
Details
Initial commit: LLM Fusion MCP Server 
- Unified access to 4 major LLM providers (Gemini, OpenAI, Anthropic, Grok)
- Real-time streaming support across all providers
- Multimodal capabilities (text, images, audio)
- Intelligent document processing with smart chunking
- Production-ready with health monitoring and error handling
- Full OpenAI ecosystem integration (Assistants, DALL-E, Whisper)
- Vector embeddings and semantic similarity
- Session-based API key management
- Built with FastMCP and modern Python tooling

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-09-05 05:47:51 -06:00

2.1 KiB
Raw Blame History

LLM Fusion MCP - Claude Code Integration Guide

Quick Setup

  1. Install the MCP server:

    ./install.sh

  2. Configure API keys in .env:

    GOOGLE_API_KEY=your_google_api_key
OPENAI_API_KEY=your_openai_api_key  # Optional
ANTHROPIC_API_KEY=your_anthropic_key  # Optional
XAI_API_KEY=your_xai_key  # Optional

  3. Add to Claude Code (recommended):

    claude mcp add -s local -- gemini-mcp /home/rpm/claude/gemini-mcp/run_server.sh

    Or via JSON configuration:

    {
  "mcpServers": {
    "gemini-mcp": {
      "command": "/home/rpm/claude/gemini-mcp/run_server.sh",
      "env": {
        "GOOGLE_API_KEY": "${GOOGLE_API_KEY}",
        "OPENAI_API_KEY": "${OPENAI_API_KEY}",
        "ANTHROPIC_API_KEY": "${ANTHROPIC_API_KEY}",
        "XAI_API_KEY": "${XAI_API_KEY}"
      }
    }
  }
}

Available Tools

🎯 Core LLM Tools

  • llm_generate() - Universal text generation across all providers
  • llm_analyze_large_file() - Intelligent large document analysis
  • llm_analyze_image() - Image understanding and analysis
  • llm_analyze_audio() - Audio transcription and analysis
  • llm_with_tools() - Function calling during generation

📊 Embeddings & Similarity

  • llm_embed_text() - Generate vector embeddings
  • llm_similarity() - Calculate semantic similarity

🔧 Provider Management

  • llm_set_provider() - Switch default provider
  • llm_get_provider() - Get current provider info
  • llm_list_providers() - List all available providers
  • llm_health_check() - Check provider status

🛠️ Utilities

  • llm_utility_calculator() - Basic math operations

Supported Providers

  • Gemini: Latest 2.5 models (up to 1M token context)
  • OpenAI: GPT-4.1, O-series reasoning models (up to 1M token context)
  • Anthropic: Claude 4 Sonnet/Haiku (200K token context)
  • Grok: Latest models (100K token context)

Testing

Test the installation:

# Test the MCP server
uvx --from . gemini-mcp

# Test all tools
uv run python test_all_tools.py