Feat/aws kendra vector search

[anatolyburtsev]

Summary

Added support for AWS Kendra as a vector store option

Features

• AWS Kendra Integration: Implements a new vector store node that connects to AWS Kendra for document indexing and retrieval
• Document Management: Support for batch document upload and deletion through Kendra's API
• Flexible Configuration: Configurable AWS region selection and credentials management
• Semantic Search: Leverages Kendra's built-in semantic search and ranking capabilities

Implementation Details

• Added new Kendra node under vector stores
• Used AmazonKendraRetriever from @langchain/aws
• Supports batch document operations via AWS SDK
• Includes proper credential handling for AWS authentication
• Compatible with existing Flowise document loaders and chains

Testing

created document store on the top of AWS Kendra and upserted Google's 10Q form from earnings call (60 pages pdf). Confirmed that related chunks are returned.
Created Rag agent from community template and replaced document store with Kendra driven one. Confirmed that agent returns relevant results

Document store configuration

Chunks page

Example of returning relevant chunks

[HenryHengZJ]

thank you!!