Index embeddings and retrieve similar items via cosine similarity with metadata filtering.
npx skills add https://github.com/copilot-ld/copilot-ld --skill libvector
Please help me install this Agent Skill. Skill: libvector Source: https://github.com/copilot-ld/copilot-ld/tree/main/packages/libvector Command: npx skills add https://github.com/copilot-ld/copilot-ld --skill libvector
libvector provides fast vector similarity search by storing embeddings with metadata and enabling cosine similarity-based retrieval for semantic search, RAG pipelines, and document similarity tasks.
Run the vector processing workflow to embed and index your documents for fast retrieval.
High-intent search queries and answers about installing and using this skill.
Semantic search finds similar documents by comparing embedding vectors using cosine similarity to calculate distance. The VectorIndex stores these embeddings with metadata, enabling fast retrieval of matching content across large document collections.
Cosine similarity enables semantic search by measuring the angle between embedding vectors to find closely aligned content. The VectorIndex applies this mathematical comparison to retrieve items with similar meaning, even when exact keywords differ.
Yes, VectorIndex supports metadata filtering alongside cosine similarity retrieval. It stores embeddings with associated metadata, allowing you to narrow down semantic search results by applying specific metadata constraints during the vector search process.
The best way to generate and index embeddings for a RAG pipeline is using a VectorProcessor to handle embedding generation and indexing. This prepares documents for fast semantic retrieval by storing vectors with metadata for cosine similarity search.
Content recommendation retrieves items by calculating cosine similarity between embedding vectors. The VectorIndex finds documents or conversations with closely matching semantic meaning, enabling threshold-based retrieval for precise content recommendations.
Yes, vector indexing supports threshold-based retrieval to control semantic search precision. By setting a cosine similarity threshold, the VectorIndex returns only items exceeding a specific similarity score, ensuring high-quality retrieval results.
gtrigGenerate 1024-dim embeddings and perform HNSW similarity searches with gobed.
akrindevGenerate text embeddings and calculate semantic similarity via the Gemini Embedding API.
AndreKuraitStore and query vector embeddings with Amazon S3 Vectors.
JhabbigImplement similarity search patterns for vector databases and ranked retrieval.
