rag-server-mcp - Based on the MCP protocol, integrate localization tools to provide a RAG service with context enhancement for LLMs.

Rag Server MCP

The MCP RAG Server is a retrieval-augmented generation service based on the Model Context Protocol. It automatically indexes project documents through local tools (ChromaDB and Ollama) and provides context enhancement capabilities for connected LLMs.

Knowledge management and memory Developer tools #Retrieval augmentation #Localization #Document indexing #MCP integration .TypeScript

rating : 2.5 points

downloads : 21

update time : 2025-04-29

What is the MCP RAG Server?

The MCP RAG Server is a server based on the Model Context Protocol (MCP). It enhances the response capabilities of connected large language models (LLMs) by indexing your project documents and providing relevant context.

How to use the MCP RAG Server?

Simply start the server and configure the client to begin using it. It supports automatic indexing of project files and provides various tools to manage documents and query context.

Applicable scenarios

Suitable for application scenarios that require enhanced generative AI capabilities, such as development tool integration, knowledge base management, and personalized content generation.

Main features

Automatic indexingAutomatically scan and index documents in the project when the server starts.

Supported file typesSupports .txt, .md, code files, .json, .jsonl, and .csv files.

MCP toolsProvides tools such as `indexDocuments`, `queryDocuments`, `removeDocument`, and others.

Advantages and limitations

Advantages

Local control: Use local models and vector storage to protect privacy.

Seamless integration: Designed specifically for the MCP ecosystem.

Intelligent context: Automatically provide relevant context.

Scalability: Built on Genkit, with more features to be added in the future.

Limitations

The initial run may be slow as it requires downloading dependencies.

Some advanced features may require additional configuration.

Limited ability to process large files.

How to use

Install Docker

Ensure that Docker Desktop or Docker Engine is installed.

Clone the code repository

Clone the code repository from GitHub and enter the directory.

Start the service

Start the server using Docker Compose.

Pull the embedding model

Ensure that the embedding model is successfully downloaded.

Usage examples

Query documentsUse queryDocuments to query relevant documents.

Remove a documentUse removeDocument to delete a specific document.

Frequently Asked Questions

What file types does the MCP RAG Server support?

How to manually index documents?

Related resources

GitHub repository

Access the project source code and documentation.

Model Context Protocol official documentation

Learn more about the features of MCP.

🚀 MCP RAG Server

The MCP RAG Server is a tool based on Model Context Protocol (MCP) and Retrieval Augmented Generation (RAG). By integrating with ChromaDB and Ollama, it offers a retrieval-augmented generation experience.

✨ Features

Automatic Indexing: Automatically index specified files when the server starts.
Filtering Support: Support excluding specific files or directories based on wildcard patterns.
Integrated Development Environment (IDE) Plugins: Support plugins for IDEs like VS Code to use the MCP service directly in the code.

🚀 Quick Start

Install Dependencies:
```
npm install
```
Build the Project:
```
npm run build
```
Start the Server:
```
npm start
```
Configure the MCP Client: In the settings of IDEs like VS Code, point the MCP service to the running RAG server.

📦 Installation

The configuration is done through environment variables, usually set in the docker-compose.yml file:

CHROMA_URL: The URL of the ChromaDB service, defaulting to http://chromadb:8000.
OLLAMA_HOST: The URL of the Ollama service, defaulting to http://ollama:11434.
INDEX_PROJECT_ON_STARTUP: Set to true or false to enable/disable automatic indexing on startup, enabled by default.
INDEXING_EXCLUDE_PATTERNS: A comma-separated list of glob patterns to exclude specific files or directories, including **/node_modules/** and **/.git/** by default.

💻 Usage Examples

Basic Usage

# Install dependencies
npm install
# Build the project
npm run build
# Start the server
npm start

Advanced Usage

# Run code linting
npm run lint
# Format the code
npm run format
# Comprehensive validation (including format check, code check, and type check)
npm run validate
# Run unit tests
npm test
# Run unit tests with coverage
npm run test:cov

📚 Documentation

The full documentation will be available at [TODO: Add link to deployed VitePress site].

🔧 Technical Details

Development

Prerequisites: Node.js (LTS version) and npm.
Install Dependencies:
```
npm install
```
Build the Project:
```
npm run build
```
Run Checkers/Formatters:
- Code Linting:
```
npm run lint
```
- Code Formatting:
```
npm run format
```
- Comprehensive Validation (including format check, code check, and type check):
```
npm run validate
```
Run Tests:
- Unit Tests:
```
npm test
```
- Unit Tests with Coverage:
```
npm run test:cov
```
Run the Server Locally (without Docker):
- Ensure ChromaDB and Ollama are running and accessible.
- Set the environment variables (CHROMA_URL, OLLAMA_HOST).
- Execute:
```
npm start
```

📄 License

This project is licensed under the MIT License.

Future Plans

Improve the code file slicing strategy.
Support more file types (e.g., PDF).
Enhance the query filtering ability.
Investigate and fix failures in end-to-end tests.
Strengthen the error handling mechanism.

Contribution

Contributions are welcome! Please discuss changes through issues before submitting a pull request. Follow the coding standards and submission specifications.