astra-db-mcp

Official
Local
38
Signed
GitHub Repo

Overview

The astra-db-mcp server is a Model Context Protocol (MCP) implementation that enables AI assistants and agents to interact directly with DataStax Astra DB — a cloud-native database service — through a standardized, agent-friendly interface. It allows natural language workflows to query, manage, and manipulate collections and records stored in Astra DB without writing custom database integration code.

Transport

stdio

Tools

  • GetCollections
  • CreateCollection
  • UpdateCollection
  • DeleteCollection
  • EstimateDocumentCount
  • ListRecords
  • GetRecord
  • CreateRecord
  • UpdateRecord
  • DeleteRecord
  • FindRecord
  • FindDistinctValues
  • BulkCreateRecords
  • VectorSearch
  • HybridSearch
  • HelpAddToClient
  • OpenBrowser

Key Capabilities

  • Full CRUD for collections and records — Create, retrieve, update, and delete data stored in Astra DB at both single-item and bulk levels.
  • Flexible querying — Find and list records using filters, estimates, and field-specific lookups.
  • Vector and hybrid search — Perform similarity search on vector embeddings and combine vector and text relevance for richer semantic queries.
  • Database introspection — Retrieve collection lists and metadata for dynamic discovery by AI workflows.
  • Batch processing — Efficiently handle large data operations with bulk create/update/delete tools.
  • Seamless AI integration — Works with MCP-aware clients like Claude Desktop, Cursor, and others to embed Astra DB capabilities directly into conversational and automated agent workflows.

How It Works

When you deploy the astra-db-mcp server, it acts as an intermediary between your MCP client and your Astra DB instance. You must provide your Astra DB Application Token and API Endpoint (and optionally a specific keyspace) as environment variables so the server can authenticate and communicate with your database.

After starting the server (commonly via npx -y @datastax/astra-db-mcp or a Docker image), your MCP client connects to it and exposes the available tools to AI agents. These tools encapsulate Astra DB operations — from managing collections to querying records and performing vector search — so AI assistants can generate natural language requests that are translated into database actions over MCP.

This design lets agents conduct database workflows (e.g., “find the latest customer support tickets with specific tags”, “insert a batch of new records”, or “perform a semantic vector search over product embeddings”) without bespoke integration layers, abstracting away API details and leveraging the MCP protocol for secure, structured interactions.