mcpField
E
embeddings

embeddings

github.com/ruvnetData & databases80.0k1.2k installsUpdated today
Repository

About embeddings

An embeddings skill for semantic search and pattern matching, aimed at workflows using claude-flow. It combines HNSW indexing, configurable text chunking, normalization, and quantization with SQLite persistence.

Key features

  1. 01HNSW-indexed semantic search
  2. 02SQLite persistent cache
  3. 03Configurable text chunking
  4. 04L1, L2, min-max and z-score normalization
  5. 05Int8, Int4 and binary quantization

Use cases

  1. 01Search document collections by meaning
  2. 02Match related patterns in memory
  3. 03Reduce embedding storage with quantization

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/ruvnet/ruflo --skill embeddings

After installing, type/embeddingsin Claude Code to confirm it loaded.

Details

From the project README

Embeddings Skill

Purpose

Vector embeddings for semantic search and pattern matching with HNSW indexing.

Features

FeatureDescription
sql.jsCross-platform SQLite persistent cache (WASM)
HNSW150x-12,500x faster search
HyperbolicPoincare ball model for hierarchical data
NormalizationL2, L1, min-max, z-score
ChunkingConfigurable overlap and size
75x fasterWith agentic-flow ONNX integration

Commands

Initialize Embeddings
npx claude-flow embeddings init --backend sqlite
Embed Text
npx claude-flow embeddings embed --text "authentication patterns"
Batch Embed
npx claude-flow embeddings batch --file documents.json
Semantic Search
npx claude-flow embeddings search --query "security best practices" --top-k 5

Memory Integration

# Store with embeddings
npx claude-flow memory store --key "pattern-1" --value "description" --embed

# Search with embeddings
npx claude-flow memory search --query "related patterns" --semantic

Quantization

TypeMemory ReductionSpeed
Int83.92xFast
Int47.84xFaster
Binary32xFastest

Best Practices

  1. Use HNSW for large pattern databases
  2. Enable quantization for memory efficiency
  3. Use hyperbolic for hierarchical relationships
  4. Normalize embeddings for consistency

Comments

Markdown supported