Logeion

Logeion is an MCP server that exposes a Latin dictionary and lemmatization functionality via the MCP protocol. It leverages spaCy's Latin model to lemmatize words and a local SQLite database (dvlg-wheel-mini.sqlite) to retrieve dictionary entries. The server runs in stdio mode by default, making it easy to connect from MCP clients and large language models. Key tools include get_word(word: str), which searches for a Latin term in the dictionary and returns the original term, its lemma if found, a list of matching results, and metadata about the search method (e.g., exact_match, lemmatized, none, or error). This setup allows LLMs to perform Latin word lookups, obtain definitions, parts of speech, lemmas, and related information in a structured, programmatic way.

Prerequisites:

• Python 3.8 or newer
• pip (or conda) for package installation

Setup steps:

1. Clone the repository:

git clone https://github.com/philipaidanbooth/logeion-mcp-server.git
cd logeion-mcp-server

2. Create and activate a virtual environment:

python -m venv venv
# macOS/Linux
source venv/bin/activate
# Windows
venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Download the spaCy Latin model:

python -m spacy download la_core_web_lg

5. Obtain and place the dictionary database in the project root:

• Place the database file named dvlg-wheel-mini.sqlite in the project root directory (as referenced by the server).

6. Run the server:

python logeion.py

Optional testing or development steps:

• Run tests:

python test_logeion.py

• Run the demo:

python demo.py

Notes and tips:

• The server uses stdio transport by default for MCP compatibility. Ensure your MCP client is configured to communicate over stdio if you’re using local testing.
• The dvlg-wheel-mini.sqlite database file must be present in the repository root for dictionary lookups to succeed. If you move the database, update the path accordingly in code or environment (if supported).
• The spaCy Latin model la_core_web_lg is required for lemmatization; ensure it is downloaded before running the server.
• If you plan to run in Docker, build with the included Dockerfile and ensure the SQLite database is mounted into the container if you choose a separate path.
• Tools and endpoints are exposed via the MCP framework; to add new tools, use the @mcp.tool() decorator in the code (as shown in the repository).
• This server is Python-based; there is no npm package associated with it (npm_package should be null).