Overview of the mcp-cps-data MCP Server

The mcp‑cps‑data server bridges the gap between conversational AI assistants and localized educational data for Chicago Public Schools. It exposes two specialized tools that let an assistant run SQL queries against a local SQLite database and retrieve contextual information from a LanceDB vector store, both populated with school‑related facts. This setup solves the common problem of integrating up‑to‑date, location‑specific educational data into AI workflows without relying on external APIs or paid services.

The server’s core value lies in its simplicity and locality. By hosting the databases locally, developers avoid network latency and privacy concerns that arise when querying public APIs. The SQLite schema contains school identifiers, names, neighborhoods, and timestamps, while the LanceDB store holds rich embeddings of school websites. This dual‑database architecture allows an assistant to answer both factual questions (“Which schools are in the Lakeview neighborhood?”) and context‑rich queries (“What is the enrollment trend for Lincoln Elementary?”) with minimal overhead.

Key capabilities include:

• – Executes arbitrary SELECT statements against the table. Developers can craft complex joins or filters to surface precise school information, while the tool handles parameter validation and error reporting.
• – Leverages LanceDB embeddings to retrieve the most relevant sections of a school’s website given a user question. Optional scoping further refines the search, enabling targeted context retrieval.

These tools are exposed through MCP’s standard resource and prompt mechanisms, making them immediately usable by any Claude or other MCP‑compliant assistant. A developer can simply add the server to their client configuration, and the assistant will automatically discover the tools during initialization.

Real‑world scenarios include:

• Educational analytics dashboards where an AI assistant walks a school administrator through enrollment statistics, budget allocations, and neighborhood demographics.
• Parent‑student portals that answer questions about nearby schools, program offerings, and safety metrics by querying the local databases in real time.
• Research assistants that pull historical school data for sociological studies, leveraging the vector store to surface relevant policy documents or news articles.

Integration is straightforward: once the server is running, an assistant can call the tools via MCP’s tool‑execution protocol. The response format is JSON, which the assistant can parse and incorporate into its natural language output. Because the data resides locally, developers have full control over updates—simply replace the SQLite or LanceDB files and restart the server, without touching code.

In summary, mcp‑cps‑data provides a lightweight, privacy‑preserving bridge between AI assistants and rich, local Chicago Public School data. Its dual database approach, coupled with MCP’s tool discovery mechanisms, empowers developers to build highly contextual, data‑driven conversational experiences tailored to educators, parents, and researchers alike.