AWS KB Retrieval
Free官方参考实现，使用 Bedrock Agent Runtime 从 AWS 知识库检索信息。
FreeFree tier
About AWS KB Retrieval
An MCP server implementation for retrieving information from the AWS Knowledge Base using the Bedrock Agent Runtime. It provides a simple tool to perform retrieval-augmented generation (RAG) queries against a specified AWS Knowledge Base, returning configurable numbers of results. The server can be run via Docker or npx and requires AWS credentials with Bedrock Agent Runtime permissions.
Key Features
Retrieval-Augmented Generation (RAG) from AWS Knowledge Base
Configurable number of results (default: 3)
Tool-based interface: retrieve_from_aws_kb
Supports Docker and npx deployment
Integrates with Claude Desktop via MCP
Pros & Cons
Pros
- Leverages AWS Bedrock Agent Runtime for secure and scalable retrieval
- Simple configuration and deployment via Docker or npx
- Customizable number of results for fine-grained control
- Integrates seamlessly with Claude Desktop and other MCP-compatible hosts
Cons
- Requires AWS account and proper credentials with Bedrock permissions
- Locked to AWS Knowledge Base – not portable to other vector stores
- Repository is archived as of May 2025, indicating no further updates
Best For
Augmenting LLM responses with enterprise knowledge from AWS Knowledge BaseBuilding question-answering systems over structured knowledge basesProviding contextual retrieval for AI assistants using Bedrock
FAQ
How do I set up AWS credentials for this server?
Obtain your AWS access key ID, secret access key, and region from the AWS Management Console. Ensure these credentials have appropriate permissions for Bedrock Agent Runtime operations.
How do I use this MCP server with Claude Desktop?
Add the server configuration to your claude_desktop_config.json, specifying either Docker or npx as the command and providing your AWS credentials as environment variables.
Can I customize the number of results retrieved?
Yes, the retrieve_from_aws_kb tool accepts an optional 'n' parameter (defaults to 3) to specify how many results to retrieve.