Claude + Ollama Fusion Execution System

A collaborative AI pipeline that combines Claude's strategic thinking with local LLM execution capabilities.

Architecture

Claude (Strategic/Planning/Review)
        ↓
Execution Proxy (Task Orchestration)
        ↓
Local LLM (Qwen/DeepSeek via Ollama)
        ↓
Result Analyzer (Claude Review + Optimization)

Project Structure

claude-ollama-fusion/
├── proxy/                  # FastAPI proxy servers
│   ├── local_proxy.py     # Lightweight proxy
│   └── enhanced_proxy.py  # Advanced proxy with task management
├── reviewer/              # Claude review and validation
│   └── review_agent.py
├── config/               # Configuration files
│   └── models.yaml
├── tasks/               # Task queue management
│   └── queue_manager.py
├── outputs/             # Generated code/results
├── logs/               # Execution logs
├── monitoring/         # System monitoring
├── templates/          # Claude prompt templates
│   ├── execute.md
│   └── review.md
└── requirements.txt    # Python dependencies

Quick Start

1. Install dependencies:

   pip install -r requirements.txt
   

2. Start Ollama with your preferred model:

   ollama pull qwen2.5-coder:32b
   ollama serve
   

3. Start the proxy server:

   uvicorn proxy.local_proxy:app --port 8800
   

4. Claude can now call:

   POST http://localhost:8800/execute
   {
     "prompt": "Write a Python quicksort algorithm",
     "model": "qwen2.5-coder:32b"
   }
   

Features

• Lightweight Proxy: Simple HTTP interface to Ollama
• Enhanced Proxy: Task tracking, file output, and result management
• Claude Review: Automated code review and validation
• Task Queues: Enterprise-grade job processing (Redis/Celery)
• Configuration Management: Flexible model and endpoint configuration
• Monitoring: Real-time system monitoring and logging