rag-architect

Design RAG pipelines with chunking, embedding tiers, ChromaDB, HyDE, and reranking.

Updated Mar 5, 2026

Search Tags:#rag #retrieval #knowledge-graph #embedding #graphrag #geosupply

Ready to run

npx skills add https://github.com/ShubhamManachekar/GeoSupply --skill rag-architect-shubhammanachekar

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: rag-architect
Source: https://github.com/ShubhamManachekar/GeoSupply/tree/main/.agent/skills/rag-architect
Command: npx skills add https://github.com/ShubhamManachekar/GeoSupply --skill rag-architect-shubhammanachekar

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Design and optimize robust RAG pipelines tailored for GeoSupply's GraphRAG, ChromaDB, and SubAgent stack, ensuring efficient retrieval, accurate grounding, and scalable evaluation across multilingual and structured data.

Core Features & Use Cases

End-to-end RAG pipeline design including chunking strategies, embedding tier mapping, retrieval, reranking, and faithfulness evaluation.
GraphRAG pattern integration for knowledge graphs and cross-entity reasoning in GeoSupply's domain.
Safety and cost controls with a fixed hallucination floor and local, zero-cost embedding/reranking components.

Quick Start

Configure a new rag-architect workflow to design a RAG pipeline for GeoSupply and run a sample retrieval and generation on a document corpus.

Frequently Asked Questions about rag-architect

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

FAQPage Schema

How do I design a RAG pipeline with ChromaDB and GraphRAG?▼

A RAG pipeline with ChromaDB and GraphRAG is designed by defining chunking strategies, mapping embedding tiers, and configuring collections for cross-entity reasoning across multilingual knowledge sources to ensure accurate grounding.

What's the best way to structure chunking strategies and embedding tiers for multilingual retrieval?▼

Structuring chunking strategies and embedding tiers for multilingual retrieval requires mapping document segments to appropriate embedding tiers and configuring local, zero-cost embedding components to support scalable evaluation across structured data.

Does GraphRAG support cross-entity reasoning for dynamic multilingual knowledge sources?▼

GraphRAG supports cross-entity reasoning for dynamic multilingual knowledge sources by integrating knowledge graph patterns to ground generation in structured entities, ensuring retrieval covers documents, entities, and dynamic data.

How do I apply HyDE query transformation and reranking in a RAG pipeline?▼

Apply HyDE query transformation and reranking in a RAG pipeline by transforming user queries into hypothetical documents and then reranking retrieved results to optimize faithful generation and meet strict faithfulness thresholds.

How do I enforce a strict 0.70 faithfulness threshold to prevent hallucinations?▼

Enforce a strict 0.70 faithfulness threshold by implementing safety controls with a fixed hallucination floor and evaluating generation outputs against retrieved context to prevent ungrounded responses in the pipeline.