jerry-ai-dev/modular-rag-mcp-server
Developer toolsThis page was translated automatically. Technical details may differ from the original. View the original
A modular RAG (Retrieval-Augmented Generation) system with MCP Server architecture. Using Skill to make AI follow each step of the spec and complete the code 100% by AI.
Overview
A pluggable, observable modular RAG (Retrieval-Augmented Generation) service framework that exposes tool interfaces externally via the MCP (Model Context Protocol) protocol, supporting direct calls by AI assistants such as Copilot / Claude. It is also a practical project and supporting teaching resource specifically designed for **. This project connects the most common core ——、、、—— of RAG interviews with currently popular application protocols into a complete, operational engineering project. 。 Thanks to the full-link pluggable architecture, you can quickly integrate it into your existing projects. No matter your background or needs, you can find the usage method that suits you. Specific usage strategies will be detailed later in "Who is suitable for this project & how to use it". - How to write (development specification documents) to drive development - How to use Spec-based automated code writing - How to use Spec for automated testing, packaging, and environment configuration - How to scale based on pluggable architectures (e.g., extending to Agent) :LLM / Embedding / Reranker / Splitter / VectorStore / Evaluator Each core component defines abstract interfaces, supports "LEGO brick-style" replacement, and allows one-click switching of the backend via configuration files with zero code modification. :BM25 Sparse Search solves precise matching of proper nouns + Dense Embedding solves synonym semantic matching. After RRF fusion, Cross-Encoder / LLM Rerank can be selected for precise sorting, balancing recall rate and accuracy rate. : Using an image-to-text strategy, Vision LLM automatically generates image descriptions and stitches them into the chunk, then reuses plain text RAG links to achieve "text search for images." : Follows the Model Context Protocol standard, can directly connect with MCP clients such as GitHub Copilot and Claude Desktop, zero front-end development, and can be used anywhere in one go.
Overview (machine translated)
A pluggable, observable modular RAG (Retrieval-Enhanced Generation) service framework, exposing tool interfaces externally via the MCP (Model Context Protocol) protocol, supporting direct calls by AI assistants such as Copilot and Claude. It is also a practical project and supporting teaching resource specifically designed for large model-related job learning and interviews.
📖 Table of Contents
- Project Overview
- Branch description
- Start quickly
- Who is suitable for this project & how to use it?
- Resume reference
- Frequently Asked Questions
- Follow-up arrangements
🏗️ Project Overview
What is this project?
This project connects the most common core steps in RAG interviews—Retrieval (Hybrid Search + Rerank), Multimodal Visual Processing (Image Captioning), RAG Evaluation (Ragas + Custom), Generation (LLM Response)—with the currently popular application protocol MCP (Model Context Protocol) into a complete, actionable engineering project.
One major highlight of the project is its easy adaptability to your own business. Thanks to the full-link pluggable architecture, you can quickly integrate it into your existing projects. No matter your background or needs, you can find the usage method that suits you. Specific usage strategies will be detailed later in “Who is suitable for this project & how to use it”.
It’s not just a project, but a whole set of ideas
More valuable than the project itself is the entire engineering approach behind it:
- How to write DEV_SPEC (development specification documents) to drive development
- How to use Skill to automatically complete code based on Spec
- How to use Skill for automated testing, packaging, and environment configuration
- How to scale based on pluggable architecture (e.g., extending to Agent)
Once you master the ideas, you can start entirely new projects and expansions yourself. Each step above is explained in a video with corresponding video explanations for each step of the above steps, so it’s recommended to watch together.
Overview of Core Competencies
| Module | Capability | Description |
|---|---|---|
| Ingestion Pipeline | PDF → Markdown → Chunk → Transform → Embedding → Upsert | Full-link data capture, supports multimodal image captioning |
| Hybrid Search | Dense (vector) + Sparse (BM25) + RRF Fusion + Rerank | A two-stage retrieval architecture with coarse recall + fine rerank |
| MCP Server | Standard MCP Protocol Exposes Tools | query_knowledge_hub、list_collections、get_document_summary |
This page was translated automatically. Technical details may differ from the original.
Install
This server does not publish a one-line install command.
Open the repository installation guide New to MCP? Start with the mcpskill.org guide →