32 lines (22 loc) · 855 Bytes

Optional CLI demo

This example runs a Python agent that connects to any OpenAI-compatible LLM provider and calls the root MCP weather server.

Setup

From the repository root:

python -m venv .venv
source .venv/bin/activate
pip install -r requirements-demo.txt
cp example.env .env

Edit .env for your provider:

OPENAI_BASE_URL=http://localhost:1234/v1   # LM Studio, Ollama, vLLM, etc.
OPENAI_API_KEY=not-needed                  # many local providers accept any value
OPENAI_MODEL=                              # optional; auto-detects first model if empty

Start your provider and load a model with tool-calling support.

Run

python examples/cli-agent/main.py "Is a trip to Hel, Poland a good idea?"

The agent uses the system prompt from prompts/travel_advisor.txt.