mcp-tool-layer

mcp-tool-layer

Enhances LLM agents with a Model Context Protocol layer for contextual reasoning and structured task execution. Includes agents that interact with semantic technologies (e.g. RDF, OWL, SPARQL), enabling hybrid symbolic-neural workflows.

Overview

Setup

1. It is recommended to use Python 3.11.
2. Docker is required.
3. Populate configs/mcp_configs.json.example, where you should replace the path placeholders to actual paths on your local machine. Make sure you remove .example from the file name to activate the settings.
4. Populate .env.example with your LLM BASE_URL and API_KEY. At least one pair of BASE_URL and API_KEY is required, local or remote.
5. Install the dependencies with pip install -e . (this installs the project in editable mode using the pyproject.toml configuration).
6. Spin up docker containers with docker compose up -d inside the docker folder.

Usage

Project structure