faiss

8.1

137

270

Facebook's library for efficient similarity search and clustering of dense vectors. Supports billions of vectors, GPU acceleration, and various index types (Flat, IVF, HNSW). Use for fast k-NN search, large-scale vector retrieval, or when you need pure similarity search without metadata. Best for high-performance applications.

vector-search

8.1

Rating

Installs

Data & Analytics

Quick Review

Excellent skill for FAISS vector search with comprehensive coverage of index types, GPU acceleration, and framework integrations. The SKILL.md provides clear code examples for all major use cases (Flat, IVF, HNSW, PQ indices), save/load operations, GPU usage, and LangChain/LlamaIndex integration. Structure is well-organized with decision guidance ('when to use'), quick start, and best practices. The performance comparison table and concrete examples make it immediately actionable for a CLI agent. Novelty is moderate-to-good: while vector search setup is straightforward, FAISS's complexity around index types, training, GPU acceleration, and parameter tuning (nprobe, ef_search, M values) makes this skill valuable for reducing token usage and preventing common mistakes. Minor room for improvement in describing the referenced index_types.md file more explicitly in overview, though as-is the skill is highly functional and useful.