EnrichMCP

Overview

EnrichMCP is a Python framework that extends the Model Context Protocol (MCP) by adding a semantic layer on top of your existing data model. Think of it as an ORM for AI agents: it turns relational tables, REST endpoints, or any custom data source into a set of typed, discoverable tools that an AI assistant can call directly. By generating Pydantic‑validated input and output schemas for each operation, EnrichMCP guarantees that every request made by an agent is well‑formed and that responses are immediately usable in downstream logic.

The core problem EnrichMCP solves is the friction between a developer’s data stack and an AI assistant’s need for structured, self‑describing APIs. Traditional approaches require hand‑written adapters or repetitive boilerplate to expose database queries, relationships, and validation rules. EnrichMCP automates this process: it introspects SQLAlchemy models or REST endpoints, extracts relationships (e.g., users → orders → products), and publishes a complete MCP service that includes schema discovery, typed tools, and relationship navigation. This eliminates the need for manual endpoint definitions and ensures that an agent’s calls remain consistent with the underlying data contract.

Key capabilities of EnrichMCP include:

• Typed Tool Generation: Every model field becomes a typed argument or return value, backed by Pydantic validation.
• Relationship Handling: One‑to‑many and many‑to‑one links are exposed as navigable properties, allowing agents to traverse data hierarchies naturally.
• Schema Discovery: Agents can query the MCP for a full description of tables, columns, and relationships, enabling dynamic exploration without prior knowledge.
• Backend Agnosticism: Whether the data lives in PostgreSQL, MySQL, a REST API, or custom logic, EnrichMCP can wrap it with minimal effort.
• Lifecycle Management: Built‑in support for async lifespans and cleanup hooks ensures that database connections or temporary resources are managed cleanly.

Real‑world use cases span e‑commerce analytics, customer support bots, and data science pipelines. For example, an AI assistant can ask “Show me all active users who have placed orders over $1,000” and receive a fully validated response by simply invoking the generated tool with appropriate filters. In a support scenario, an agent can navigate from a ticket to the associated customer record and retrieve recent purchase history without any custom code.

EnrichMCP integrates seamlessly into existing AI workflows by presenting a standard MCP endpoint. A developer can spin up the service, expose it over HTTP or a local socket, and then configure an AI assistant to use that endpoint as its tool source. Because the schema is part of the MCP contract, agents automatically gain up‑to‑date knowledge of the data model without manual reconfiguration. This tight coupling between data and AI logic reduces maintenance overhead, speeds up feature iteration, and ensures that agents always interact with a trustworthy, validated API surface.