Zrald-1 MCP Server

Advanced Graph RAG MCP Server with file location identification, graph processing, and result summarization capabilities.

Features

• File Location Identification: Intelligently identify and locate files based on user queries
• Graph Processing: Convert files into knowledge graph representations with nodes and relationships
• Vector Search: Semantic search across file content using vector embeddings
• Relationship Analysis: Analyze relationships between files based on content similarity, references, and dependencies
• Comprehensive Summarization: Generate detailed summaries and analytics of processed files
• Export Capabilities: Export graph data in multiple formats (JSON, Cypher, GraphML)

Installation

npm install zrald-1

Usage

As a standalone MCP server

npx zrald-1

As a library

import { GraphRAGMCPServer } from 'zrald-1';

const server = new GraphRAGMCPServer();
await server.initialize();
await server.start();

Available Tools

File Processing Tools

1. identify_files - Identify and locate files based on search criteria
2. process_files_to_graph - Convert identified files into graph nodes and relationships
3. generate_file_summary - Generate comprehensive summaries of processed files
4. search_file_content - Search within file content using vector similarity
5. analyze_file_relationships - Analyze relationships between files
6. export_graph_data - Export graph data in various formats

Vector Search Tools

7. vdb_search - Vector similarity search across all processed content

Analytics Tools

8. graph_analytics - Get comprehensive analytics about the file graph

Configuration

Create a .env file in your project root:

VECTOR_DIMENSION=384
MAX_VECTOR_ELEMENTS=10000

Tool Examples

Identify Files

{
  "name": "identify_files",
  "arguments": {
    "query": "typescript configuration",
    "search_paths": ["./src", "./config"],
    "file_types": [".ts", ".json"],
    "recursive": true,
    "max_results": 20
  }
}

Process Files to Graph

{
  "name": "process_files_to_graph",
  "arguments": {
    "file_ids": ["file-id-1", "file-id-2"],
    "create_chunks": true,
    "chunk_size": 1000,
    "chunk_overlap": 100
  }
}

Generate Summary

{
  "name": "generate_file_summary",
  "arguments": {
    "include_content_analysis": true,
    "include_relationships": true,
    "include_statistics": true,
    "summary_type": "comprehensive"
  }
}

Search File Content

{
  "name": "search_file_content",
  "arguments": {
    "query": "database connection configuration",
    "top_k": 10,
    "similarity_threshold": 0.7,
    "file_types": [".js", ".ts", ".json"]
  }
}

Resources

The server provides access to several resources:

• files://processed-files - All processed files and metadata
• graph://file-graph - Knowledge graph representation
• analytics://file-analytics - Analytics and statistics

Development

Build

npm run build

Development Mode

npm run dev

Testing

npm test

Architecture

The server consists of several key components:

• FileProcessor: Handles file identification, processing, and content analysis
• VectorStore: Manages vector embeddings and similarity search
• GraphRAGMCPServer: Main MCP server implementation with tool handlers

License

MIT

Contributing

Contributions are welcome! Please read the contributing guidelines before submitting PRs.

Support

For issues and questions, please use the GitHub issue tracker.