DataStax Astra DB MCP Server – Vector Database Tools
The DataStax Astra DB MCP Server lets MCP clients interact with Astra DB Serverless databases from natural language workflows. It is useful when an AI agent needs to inspect collections, manage records, create vector-enabled collections, or run vector and hybrid searches against an Astra DB database.
Overview
The DataStax Astra DB MCP Server is an official MCP server for connecting AI agents and MCP-compatible clients to Astra DB Serverless. It gives an agent a controlled way to work with an Astra DB database using the same application token and API endpoint that a normal application would use. DataStax documents the integration for clients such as Claude Desktop and Cursor, while the server package is published as @datastax/astra-db-mcp and maintained in the DataStax GitHub organization.
What the MCP server enables
The server exposes Astra DB workload tools for collection management, record operations, bulk record operations, vector search, hybrid search, and setup assistance. Documented tools include GetCollections, CreateCollection, UpdateCollection, DeleteCollection, EstimateDocumentCount, ListRecords, GetRecord, CreateRecord, UpdateRecord, DeleteRecord, FindRecord, FindDistinctValues, BulkCreateRecords, BulkUpdateRecords, BulkDeleteRecords, VectorSearch, and HybridSearch.
These tools allow an AI agent to inspect database structure, create or update collections, add or retrieve records, perform bulk changes, and search embeddings. CreateCollection supports vector configuration, including enabling vector search, choosing a vector dimension, and selecting a similarity metric. VectorSearch performs similarity search over embeddings, while HybridSearch combines vector similarity with text search.
When to use it
Use this MCP server when an agent needs direct, live access to Astra DB Serverless for database administration, prototyping, RAG data inspection, vector collection setup, or data maintenance tasks. Practical examples include asking an agent to create a vector-enabled collection, check whether records exist, update test data, run a similarity search over stored embeddings, or verify database changes in the Astra Portal.
Connection and authentication
The documented local connection is a command-based MCP server started with npx -y @datastax/astra-db-mcp, which means the transport represented here is local stdio. Authentication is configured with environment variables. The server requires ASTRA_DB_APPLICATION_TOKEN and ASTRA_DB_API_ENDPOINT; ASTRA_DB_KEYSPACE is optional when a specific keyspace is required instead of the default configured keyspace.
Key considerations
The token used with this server determines what the agent can do. DataStax documentation calls for an application token with the Database Administrator role, so teams should limit access to trusted users and environments, store tokens as secrets, and review tool calls before approving destructive operations. The server includes create, update, delete, and bulk delete tools, so MCP clients that ask for permission before running tools should keep that approval workflow enabled. No hosted HTTP or SSE endpoint is documented for this server in the official setup instructions, so unsupported transports are intentionally omitted.
Supported Transports
stdio
Command: npx
Args:
-y@datastax/astra-db-mcp
Frequently Asked Questions
- When should an AI agent use the DataStax Astra DB MCP Server?
- Use it when an agent needs to work with an Astra DB Serverless database from an MCP client, such as creating collections, inspecting records, updating data, testing RAG data workflows, or running vector and hybrid searches over stored embeddings.
- What does this MCP server add to an AI agent's capabilities?
- It gives the agent live Astra DB tools instead of relying on static knowledge. The agent can call documented tools for collection management, record CRUD operations, bulk record changes, vector similarity search, hybrid search, and setup assistance.
- What can the AI agent access or manage through this server?
- The agent can access the Astra DB database identified by `ASTRA_DB_API_ENDPOINT` and authorized by `ASTRA_DB_APPLICATION_TOKEN`. Within that scope, it can list and manage collections, read and modify records, perform bulk operations, and run vector or hybrid search tools. Access is limited by the supplied token and target endpoint.
- How is authentication configured for this MCP server?
- Authentication is configured through environment variables. The official README uses `ASTRA_DB_APPLICATION_TOKEN` for the Astra DB application token and `ASTRA_DB_API_ENDPOINT` for the database API endpoint. `ASTRA_DB_KEYSPACE` can be added when a specific keyspace is required. Secrets should be stored in the MCP client's secure configuration and never hard-coded into shared files.
- Which transport should be used for this MCP server?
- Use the local command-based `stdio` transport documented for Claude Desktop and Cursor. The server is started with `npx -y @datastax/astra-db-mcp`. Official setup instructions do not document a hosted Streamable HTTP or SSE endpoint for this server, so those transports are not included.