Claude Flow Embeddings

Claude Flow Embeddings

V3 Embedding Service supporting OpenAI, Transformers.js, and ONNX providers with HNSW indexing, semantic search, hyperbolic embeddings, and neural substrate integration.

Quick Command Reference

Core Commands

embeddings init

Initialize embedding subsystem with ONNX model.

npx @claude-flow/cli@latest embeddings init

embeddings generate

Generate embeddings for text.

npx @claude-flow/cli@latest embeddings generate --text "query text"

embeddings search

Semantic similarity search across stored embeddings.

npx @claude-flow/cli@latest embeddings search --query "search pattern"

embeddings compare

Compare similarity between texts.

npx @claude-flow/cli@latest embeddings compare --texts "text a" "text b"

embeddings collections

Manage embedding collections (namespaces).

npx @claude-flow/cli@latest embeddings collections

embeddings index

Manage HNSW indexes for fast nearest-neighbor search.

npx @claude-flow/cli@latest embeddings index

embeddings providers

List available embedding providers.

npx @claude-flow/cli@latest embeddings providers

embeddings chunk

Chunk text for embedding with overlap.

npx @claude-flow/cli@latest embeddings chunk --text "long document text"

embeddings normalize

Normalize embedding vectors.

npx @claude-flow/cli@latest embeddings normalize

embeddings hyperbolic

Hyperbolic embedding operations (Poincare ball model).

npx @claude-flow/cli@latest embeddings hyperbolic

embeddings neural

Neural substrate features (RuVector integration).

npx @claude-flow/cli@latest embeddings neural

embeddings models

List and download embedding models.

npx @claude-flow/cli@latest embeddings models

embeddings cache

Manage embedding cache.

npx @claude-flow/cli@latest embeddings cache

embeddings warmup

Preload embedding model for faster first inference.

npx @claude-flow/cli@latest embeddings warmup

embeddings benchmark

Run embedding performance benchmarks.

npx @claude-flow/cli@latest embeddings benchmark

Common Patterns

Initialize and Generate Embeddings

# Initialize with default ONNX model
npx @claude-flow/cli@latest embeddings init

# Warmup model
npx @claude-flow/cli@latest embeddings warmup

# Generate embedding
npx @claude-flow/cli@latest embeddings generate --text "authentication pattern"

Semantic Search Workflow

# Search stored embeddings
npx @claude-flow/cli@latest embeddings search --query "error handling best practices"

# Compare similarity of two texts
npx @claude-flow/cli@latest embeddings compare --texts "JWT auth" "OAuth2 flow"

Benchmark Embedding Performance

npx @claude-flow/cli@latest embeddings benchmark

Key Options

• --text: Input text for embedding generation
• --query: Search query for semantic search
• --texts: Multiple texts for comparison
• --verbose: Enable verbose output
• --format: Output format (text, json, table)

Programmatic API

import { EmbeddingService, HNSWIndex } from '@claude-flow/embeddings';

// Initialize with ONNX provider
const embeddings = new EmbeddingService({ provider: 'onnx' });
await embeddings.init();

// Generate embeddings
const vector = await embeddings.generate('query text');

// Semantic search
const results = await embeddings.search('pattern', { limit: 10 });

// Compare similarity
const similarity = await embeddings.compare('text a', 'text b');

RAN DDD Context

Bounded Context: Cross-Cutting Related Skills: claude-flow-memory, claude-flow-neural

References

• Complete command reference: See references/commands.md
• Full README
• npm