Skill UI

Provides a systematic guide for diagnosing and improving poor search quality in vector databases like Qdrant. This guide covers troubleshooting issues such as low recall, when approximate search degrades, selecting optimal embedding models, and correctly tuning HNSW parameters, quantization, and filtering strategies. Essential for optimizing RAG and semantic search pipelines.

A spectral vector search tool that enhances nearest-neighbor search with graph Laplacian features, capturing both semantic similarity and structural roles in embedding spaces.

This tool ingests raw market data (OHLCV) for a specified symbol. It normalizes the data into relative OHLCV vectors, converts these vectors into high-dimensional embeddings, and then stores and indexes them using HNSW. This process prepares the data for fast, similarity-based pattern detection and advanced financial time series analysis.

This skill helps discover natural groupings within a large collection of semantic embeddings using the `ruvector` tool. It employs graph-based algorithms (Spectral/Louvain) to cluster code or data, revealing underlying themes, identifying outliers, and organizing vast vector sets. Ideal for analyzing codebases or large datasets to understand relationships between entries.

This skill utilizes hyperbolic space embedding (Poincare ball model) to represent complex data structures with inherent hierarchy, such as dependency trees, ontologies, or module taxonomies. Unlike standard Euclidean embeddings, hyperbolic space preserves hierarchical distances efficiently, allowing for accurate distance calculations (geodesic distance) that grow logarithmically with tree depth. It is ideal for advanced knowledge graph analysis and code architecture mapping.

This plugin provides advanced vector search capabilities using two distinct paths. The large-scale path supports millions of vectors via HNSW, ideal for general corpus search. The hot-path router handles small, critical pattern routing (≤11 patterns) with sub-ms latency. Key features include RaBitQ 1-bit quantization for memory efficiency (32× reduction), hyperbolic space search for hierarchical data, and unified cross-namespace search.