Embedding Search

Vector-based semantic search using simulated embeddings. Provides hybrid search combining keyword matching with semantic similarity.

Capabilities

• Semantic document search
• Vector similarity matching
• Document chunking and indexing
• Hybrid keyword + vector search
• Relevance scoring and ranking
• Full-text search fallback
• Document categorization
• Configurable similarity thresholds

When to Use

Use the embedding-search skill when:

• Searching through large document collections
• Need semantic similarity matching
• Building a knowledge base
• Finding related documents
• Implementing RAG (Retrieval Augmented Generation)

Usage Examples

Index documents

node /job/.pi/skills/embedding-search/embed.js index /path/to/documents --output index.json

Search documents

node /job/.pi/skills/embedding-search/embed.js search "machine learning concepts" --index index.json

Interactive mode

node /job/.pi/skills/embedding-search/embed.js --interactive --index index.json

Add documents to existing index

node /job/.pi/skills/embedding-search/embed.js add new-doc.md --index index.json

Hybrid search with weights

node /job/.pi/skills/embedding-search/embed.js search "cloud deployment" --hybrid --keyword-weight 0.3 --semantic-weight 0.7