Hybrid Search Architect

Design hybrid retrieval pipelines combining dense vector search with BM25 keyword matching.

9|1|Updated Mar 15, 2026

Search Tags:#rag #retrieval #architect #vector-search #hybrid #bm25 #reciprocal-rank-fusion

Ready to run

npx skills add https://github.com/Notysoty/openagentskills --skill hybrid-search-architect

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: Hybrid Search Architect
Source: https://github.com/Notysoty/openagentskills/tree/main/skills/hybrid-search-architect
Command: npx skills add https://github.com/Notysoty/openagentskills --skill hybrid-search-architect

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Designs a hybrid retrieval pipeline that combines dense vector search (semantic similarity) with BM25 sparse search (keyword matching). Hybrid search outperforms either method alone on most retrieval benchmarks because vector search handles semantic meaning while BM25 handles exact keyword matches, product names, codes, and rare terms. This skill picks the right combination and fusion strategy for your use case.

Core Features & Use Cases

Architectural guidance for selecting a hybrid vs vector-only or BM25-only approach.
Step-by-step configuration for reciprocal rank fusion and stack tuning across popular backends (Weaviate, Elasticsearch/OpenSearch, pgvector).
Real-world scenarios include product documentation search, code search, and knowledge bases with mixed content.

Quick Start

Copy this file to .agents/skills/hybrid-search-architect/SKILL.md in your project root to enable the hybrid retrieval architecture.

Frequently Asked Questions about Hybrid Search Architect

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

What is hybrid search and when do I need it for my retrieval pipeline?▼

Hybrid search combines dense vector similarity for semantic meaning with BM25 keyword matching for exact terms. You need it when queries mix semantic intent with exact keywords like product codes, outperforming either method alone on retrieval benchmarks.

How do I configure reciprocal rank fusion for hybrid search?▼

Configuring reciprocal rank fusion involves merging result lists from dense vector and BM25 sparse components. This skill provides step-by-step configuration guidance for tuning fusion parameters across backends like Weaviate, Elasticsearch, and pgvector.

Does hybrid search work with pgvector and Elasticsearch backends?▼

Yes, hybrid search supports multiple backends including pgvector, Elasticsearch, OpenSearch, and Weaviate. You can configure the retrieval pipeline and reciprocal rank fusion across these popular search databases to match your existing infrastructure.

What is the best way to improve retrieval precision for mixed content queries?▼

The best way to improve retrieval precision for mixed content is using a hybrid search pipeline with optional reranking. Fusing dense vector and BM25 results via reciprocal rank fusion captures both semantic meaning and exact keyword matches effectively.

When should I not use hybrid search and rely on vector-only retrieval?▼

Rely on vector-only retrieval instead of hybrid search when your dataset contains purely semantic content without exact keywords, product names, or rare terms. Hybrid search is unnecessary if BM25 keyword matching provides no distinct value for your queries.