Glue MCP – Connecting AI Assistants to the AWS Glue Data Catalog

Glue MCP is a lightweight Model Context Protocol server that exposes the metadata of an AWS Glue Data Catalog as a set of AI‑ready tools. Developers can run the server locally and have their Claude or Ollama models discover, query, and manipulate Glue databases and tables through a simple, well‑defined MCP interface. The primary problem it solves is the friction that normally accompanies integrating cloud data catalogs into generative‑AI workflows: without Glue MCP, an assistant would need custom SDK calls or REST wrappers for every operation, leading to brittle code and a steep learning curve.

At its core, Glue MCP listens on an HTTP port (default 8000) and implements the standard MCP protocol over Server‑Sent Events. Once a client connects, it automatically publishes three tools: , , and . These tools translate natural‑language queries into Glue API calls, returning JSON metadata that the assistant can interpret or pass to downstream functions. By encapsulating Glue’s API behind a uniform tool interface, developers can focus on higher‑level logic—such as data exploration or automated report generation—without handling authentication, pagination, or error handling for each request.

Key capabilities include:

• Seamless credential integration – the server inherits AWS credentials from the environment, so there is no need to hard‑code secrets or manage IAM roles in application code.
• SSE‑based streaming – clients receive incremental responses, enabling real‑time feedback and smoother user experiences.
• Zero‑configuration client setup – a simple JSON configuration file lets any MCP‑compliant tool host (e.g., ) register the server with a friendly name, making it discoverable by AI assistants instantly.

Typical use cases are plentiful. A data analyst can ask an assistant, “Show me all databases in Glue,” and receive a concise list; a data engineer can request table schemas to auto‑generate ETL pipelines; or an executive might ask for a summary of data lineage across databases. In automated ML pipelines, the assistant can invoke Glue MCP to pull table metadata and feed it into training data discovery scripts, all within the same conversational flow.

Glue MCP’s standout advantage lies in its minimalism and adherence to MCP standards. It requires no external dependencies beyond AWS credentials, making it trivial to spin up in CI/CD pipelines or local dev environments. Because the server exposes only a curated set of tools, it reduces attack surface and simplifies auditing. For developers already comfortable with MCP concepts, Glue MCP provides a plug‑and‑play bridge to one of the most widely used cloud data catalogs, unlocking powerful AI‑driven data discovery and management workflows.