chris-assistant 0.1.0

An AI-powered CLI for exploring the ChRIS medical image platform using MCP and Llama3

🧠 ChRIS-Assistant

An AI-powered CLI assistant for exploring the ChRIS medical image platform using natural language.

This tool connects to a running ChRIS MCP Server, dynamically fetches its tools, and uses Ollama with Llama3 to route questions and summarize responses.

---

🚀 Features

• 🔌 Connects to any MCP-compatible ChRIS server over SSE
• 🤖 Uses llama3 via Ollama to understand natural questions
• 🧠 Automatically selects the right ChRIS API tool
• 🧾 Summarizes JSON output into plain English
• 🧰 Built with llama-index + mcp

---

📦 Installation

1. Clone and install locally (dev mode)

git clone https://github.com/YOUR_GITHUB/chris-assistant.git
cd chris-assistant
pip install -e .

2. Or install via PyPI (once published)

pip install chris-assistant

---

💬 Usage

Make sure you have Ollama running and the ChRIS MCP server accessible:

OLLAMA_HOST=localhost:11434 \
MCP_URL=http://localhost:3001/sse \
LLM_MODEL=llama3.2 \
chris-assistant

---

💡 Example Questions

What's at the root of the ChRIS API?
List available plugins
Show me plugin instance 2

---

⚙️ Environment Variables

Variable	Description	Default
MCP_URL	MCP server SSE endpoint	http://localhost:3001/sse
OLLAMA_HOST	Ollama runtime (local model endpoint)	localhost:11434
LLM_MODEL	LLM model name (e.g. llama3.2)	llama3.2

---

📚 Project Layout

chris_assistant/
├── chris_assistant/
│   └── cli.py        # Main CLI logic
├── pyproject.toml    # Packaging config
├── README.md

---

📜 License

MIT © [Your Name]

---