About This Architecture

Hybrid retrieval architecture combining Claude and Codex coding agents with an MCP server gateway that orchestrates both keyword and semantic search through FTS5 and sqlite-vec. The system routes queries through a Session Tracker and dual-path retrieval pipeline—FTS5 for keyword matching and sqlite-vec for vector similarity—then fuses results via Reciprocal Rank Fusion before applying provenance tracking and promotion workflows. SQLite persistence with WAL mode ensures concurrent access across local and remote machines via Cross-Machine Bridge Sync, while Auth/Access Control and Prompt/Query Cache enforce security and reduce redundant computation. Fork this diagram to customize the retrieval fusion strategy, add vector embedding models, or extend the promotion workflow for your knowledge base or RAG application.

People also ask

How do you build a hybrid retrieval system that combines keyword and semantic search with Claude agents and SQLite?

This diagram shows a complete hybrid retrieval architecture where Claude and Codex agents query an MCP server gateway that routes requests through parallel FTS5 keyword search and sqlite-vec semantic search pipelines, fuses results via Reciprocal Rank Fusion, and persists data in SQLite with WAL mode for safe concurrent access across local and remote machines.

Sqlite Memory MCP Hybrid Retrieval Architecture

Autoadvancedhybrid-retrievalsqliteMCPsemantic-searchRAGClaude
Domain: Software ArchitectureAudience: AI engineers and full-stack developers building hybrid retrieval systems with Claude and MCP servers
0 views0 favoritesPublic

Created by

July 24, 2026

Updated

July 24, 2026 at 1:27 PM

Type

architecture

Need a custom architecture diagram?

Describe your architecture in plain English and get a production-ready Draw.io diagram in seconds. Works for AWS, Azure, GCP, Kubernetes, and more.

Generate with AI

AI-generated. Verify before production use. Learn more

Report this diagram