About Fastmcp Chroma Ollama
Fastmcp Chroma Ollama is an MCP server in the RAG category: local memory system for LLM assistants using ChromaDB, Ollama, and FastMCP.
Install
git clone https://github.com/bjkemp/fastmcp-chroma-ollamaThis server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.
Ask AI
Ask AI about Fastmcp Chroma Ollama
Powered by Claude · Grounded in docs
Security checks
- ✓README presentThe repository ships documentation.
5 checks have not been run yet.
README
FastMCP ChromaDB Memory Server
Overview
A sophisticated, local memory system for AI assistants using ChromaDB, Ollama, and FastMCP. This project provides a persistent, semantically-aware memory storage and retrieval system with advanced features like memory importance scoring, automatic merging, and intelligent pruning.
🌟 Features
- 🧠 Semantic Memory Storage
- 🔒 Local-first Architecture
- 🚀 High-Performance Vector Search
- 🤖 Ollama Embedding Support
- 📊 Memory Importance Scoring
- 🔍 Advanced Retrieval Mechanisms
🛠 Requirements
- Python 3.10+
- UV Package Manager
- Ollama
- ChromaDB
- FastMCP
📦 Installation
1. Install UV Package Manager
# macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh
# Windows
powershell -c "irm https://astral.sh/uv/install.ps1 | iex"
2. Install Ollama
curl https://ollama.ai/install.sh | sh
ollama pull nomic-embed-text
3. Install Project
# Clone the repository
git clone https://github.com/bjkemp/fastmcp-chroma-ollama.git
cd fastmcp-chroma-ollama
# Create virtual environment
uv venv
# Activate virtual environment
source .venv/bin/activate # On Windows: .venv\Scripts\activate
# Install project
uv pip install -e .
🚀 Usage
Starting the Server
# With stdio transport (for Claude Desktop)
uv run -m fastmcp_memory.server --transport stdio
# With SSE transport
uv run -m fastmcp_memory.server --transport sse --port 8080
Basic Python Example
from fastmcp_memory import MemoryManager
async def example():
manager = MemoryManager()
# Store a memory
memory_id = await manager.add_memory("The capital of France is Paris")
# Retrieve memories
memories = await manager.retrieve_memories("France's capital")
print(memories)
🧪 Development
Setup Development Environment
# Install development dependencies
uv pip install .[dev]
# Run tests
uv run pytest
# Run code formatters
uv run black src tests
uv run isort src tests
📋 Project Structure
fastmcp-chroma-ollama/
├── src/
│ └── fastmcp_memory/
│ ├── __init__.py
│ ├── server.py
│ ├── server_core.py
│ └── ...
├── tests/
├── .github/
│ └── workflows/
├── pyproject.toml
├── README.md
├── TODO.md
└── LICENSE
🤝 Contributing
Contributions are welcome! Please check our TODO.md for current development priorities and see our contribution guidelines.
📄 License
MIT License - See LICENSE for details.
🗺 Roadmap
Check out TODO.md for detailed development plans and future directions.
🙏 Acknowledgements
README copy captured when this server was indexed; it may lag the repository. The original is authoritative.