AWS KB
Source Code
Description
AWS KB is an MCP server that enables users to retrieve information from AWS Knowledge Bases using Retrieval-Augmented Generation (RAG) via the Bedrock Agent Runtime. It allows querying AWS Knowledge Bases and returns both raw context and structured RAG sources.
Features
- Retrieval-Augmented Generation (RAG): Retrieve context from AWS Knowledge Bases based on a query and Knowledge Base ID.
- Multiple Results Retrieval: Option to retrieve a customizable number of results (default is 3).
- Flexible Output: Returns both raw context (text) and structured results (JSON) with metadata (id, fileName, snippet, score).
- AWS Credentials Support:
- Supports configuration using IAM Access Keys (with optional session token for temporary credentials).
- Supports AWS SSO (Single Sign-On) credentials.
- Default Knowledge Base IDs:
- Optionally specify one or more default Knowledge Base IDs via environment variable.
- Easy Integration:
- Can be run via Docker, NPX, or directly from a cloned/built repository.
- Example configurations for integration with Claude Desktop and other MCP-compatible systems.
- Open Source: Licensed under the MIT License.
Usage
- Configure AWS credentials using either IAM Access Keys or AWS SSO.
- Specify Knowledge Base IDs as needed.
- Run the server using Docker, NPX, or Node.js.
- Supports flexible integration and deployment options.
Pricing
- Free and Open Source: Licensed under the MIT License; no pricing plans.
aws rag knowledge-base bedrock