How does this skill improve search speed?

It uses HNSW (Hierarchical Navigable Small World) indexing and ONNX integration to achieve search speeds significantly faster than standard linear comparisons.

Can I use this for hierarchical data?

Yes, the skill includes Hyperbolic support using the Poincare ball model, which is specifically designed to represent and search hierarchical relationships effectively.

What is the benefit of sql.js in this skill?

sql.js provides a persistent, cross-platform SQLite cache for embeddings using WebAssembly, ensuring that your indexed data remains available across different sessions.

Does it support quantization to save memory?

Yes, it supports Int8, Int4, and Binary quantization, allowing for up to a 32x reduction in memory usage for large-scale vector databases.

Semantic Vector Embeddings

Name: Semantic Vector Embeddings
Author: nbossn

bynbossn

0•

Data Science & ML

Empowers Claude with high-performance vector embeddings, HNSW indexing, and persistent semantic search capabilities.

The Embeddings skill integrates advanced vector search capabilities into the Claude Code environment, enabling high-speed semantic retrieval and pattern matching. By utilizing HNSW indexing and sql.js for persistent caching, it achieves significant performance gains compared to standard implementations. The skill supports sophisticated features like hyperbolic models for hierarchical data, various normalization techniques, and quantization for memory efficiency, making it an essential tool for large-scale knowledge retrieval and complex similarity queries within agentic workflows.

Key Features

01Memory Quantization: Includes Int8, Int4, and Binary options to reduce memory footprint by up to 32x.

02Agentic Integration: Optimized for agentic-flow with ONNX support for 75x faster execution.

03HNSW Indexing: Accelerates semantic search by 150x to 12,500x compared to linear scans.

04Persistent Storage: Utilizes sql.js (WASM) for cross-platform SQLite-based embedding caching.

050 GitHub stars

06Hyperbolic Support: Implements Poincare ball models for representing hierarchical data relationships.

Use Cases

01Hierarchical Mapping: Representing complex file structures or class hierarchies using hyperbolic embeddings.

02Pattern Matching: Identifying similar code structures or architectural patterns across repositories.

03Knowledge Retrieval: Building RAG-based systems for searching massive documentation and codebases.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add nbossn/claude-code-base embeddings

For use in Claude.ai and ChatGPT