Amazon Kendra Index MCP Server

What is Amazon Kendra Index MCP Server?

Amazon Kendra Index MCP Server is an MCP-compliant server that exposes capabilities of Amazon Kendra indexes to AI-powered applications and agents. It allows users and LLMs to list available Kendra indexes and perform semantic search queries against indexed enterprise data using natural language, providing relevant documents and insights as context for generative AI tasks and workflows.

How to Configure Amazon Kendra Index MCP Server

1. Prerequisites:

   • Ensure you have an AWS account with Amazon Kendra enabled.
   • Set up appropriate IAM permissions so the server can access Kendra indexes and perform search/query operations.
   • Install Python 3.10+ and uv.

2. Installation:

   • Install the Amazon Kendra Index MCP Server using:

     uvx awslabs.amazon-kendra-index-mcp-server@latest
     

   • Alternatively, include this server in your MCP client configuration file (e.g., mcp.json, cline_mcp_settings.json) like:

     {
       "awslabs.amazon-kendra-index-mcp-server": {
         "command": "uvx",
         "args": [
           "awslabs.amazon-kendra-index-mcp-server@latest"
         ],
         "env": {
           "AWS_PROFILE": "your-aws-profile",
           "AWS_REGION": "us-east-1",
           "FASTMCP_LOG_LEVEL": "ERROR"
         }
       }
     }
     

   • Adjust AWS_PROFILE and AWS_REGION as needed for your environment.

3. Credentials:

   • Make sure AWS credentials (profile, environment variables, or EC2 roles) are configured to permit access to Amazon Kendra resources.

4. (Optional) Docker Support:

   • You can build and run the server as a Docker container if you prefer containerized deployments.

How to Use Amazon Kendra Index MCP Server

1. Enable the Server: After configuring, start your MCP client or agent (such as Cline, Cursor, Windsurf, etc.) with the server listed as enabled.

2. Discover Kendra Indexes: You can ask your AI assistant, or invoke tools directly, to list available Kendra indexes in your account. For example:

   • "List all available Amazon Kendra indexes."

3. Query Kendra Indexes: Use natural language to search across the indexed enterprise data. For example:

   • "Find all documents about our data security policy using the main corporate Kendra index."

4. Result Handling: The MCP client or LLM agent will use the relevant search results to enhance its understanding, suggest actions, or generate responses leveraging up-to-date, domain-specific knowledge from your indexed sources.

5. Integration with Agents: LLM-based agents can integrate Kendra search results automatically in RAG (Retrieval-Augmented Generation) workflows, improving answer factuality and context.

Key Features

• Index Listing: Enumerate all available Amazon Kendra indexes in your AWS account.
• Natural Language Semantic Search: Issue open-ended or targeted queries to Kendra indexes using plain English, retrieving the most relevant documents or passages.
• Enterprise RAG Context: Enable generative AI assistants to tap into structured organizational knowledge for more accurate, context-rich outputs.
• Secure and Configurable: Leverages AWS authentication for secure data access.
• Flexible Integration: Works seamlessly with any MCP-enabled client, agent, or development environment.

Use Cases

• AI-powered Knowledge Search: Let AI chatbots and assistants answer user questions with information pulled from enterprise documentation, wikis, manuals, or technical repositories indexed in Kendra.
• Developer Assistant Context Injection: Provide coding assistants and IDE agents with authoritative, up-to-date guidance, snippets, or policy documents relevant to the developer's project or query.
• Automated Compliance Checks: Retrieve and present policy documents, compliance checklists, or instructions in response to audit or regulatory queries.
• Enterprise Q&A Enhancement: Use with LLMs in customer-support or HR tools to answer employee questions using trusted organizational sources.
• RAG (Retrieval-Augmented Generation) Pipelines: Fetch precise context from Kendra to ground generative AI responses in factual company data.

Frequently Asked Questions

Q1: Do I need an Amazon Kendra subscription to use this server?
Yes, you must have Amazon Kendra provisioned in your AWS account and one or more indexes created.

Q2: Can I restrict which indexes are exposed to AI agents?
Yes, access is controlled via AWS IAM permissions. You can scope allowed indexes by resource-based policies.

Q3: What data can AI assistants access through the Kendra Index MCP Server?
Only the data indexed in your configured Kendra indexes and to which the server's IAM credentials permit access.

Q4: Are queries to Kendra in real-time or cached?
Queries are live—each operation fetches the latest available results from Kendra (subject to Kendra's indexing schedule).

Q5: How are large result sets handled?
By default, only the top relevant documents or snippets are returned (pagination or result limits may apply, based on configuration and tool parameters).