vector-databases

What problem does it solve?

This Skill enables the creation and management of vector databases for efficient similarity search, handling complex retrieval scenarios like filtering and hybrid search.

Core Features & Use Cases

• Vector Ingestion: Store data as vectors with associated metadata payloads.
• Filtered Search: Perform similarity searches that respect metadata constraints.
• Hybrid Retrieval: Combine multiple search signals (e.g., dense and sparse vectors) for improved accuracy.
• Index Maintenance: Optimize database performance as data changes.
• Use Case: When building a RAG system, use this skill to ingest document chunks with metadata, perform filtered searches based on user permissions, and combine dense embedding search with keyword search for more relevant results.

Quick Start

Use the vector-databases skill to build a point with id 'doc1', a vector '[0.1, 0.2]', and payload '{"source": "web"}'