NLWeb Retrieval Backend Manager
Configures and manages diverse retrieval backends for NLWeb, enabling seamless integration with vector stores and search engines.
This skill facilitates the selection, configuration, and debugging of various retrieval endpoints within the NLWeb ecosystem, including local Qdrant, cloud-based Azure AI Search, and database extensions like pgvector. It guides developers through the unique 'read-fanout' architecture where queries are executed across multiple enabled backends in parallel while writes are directed to a single canonical store. By providing detailed setup instructions for specific providers and ensuring embedding dimension compatibility, it simplifies the process of building robust, multi-source RAG pipelines for agentic commerce applications.