C++ Graph-RAG MCP Server

C++ Graph-RAG MCP Server

A Model Context Protocol server for analyzing large C++ codebases with semantic search, crash dump analysis, and a web UI for configuration.

Category
Visit Server

README

C++ Graph-RAG MCP Server

A powerful Model Context Protocol (MCP) server for analyzing large C++ codebases using Graph-RAG architecture. Features semantic search, crash dump analysis, and a web UI for configuration.

Features

  • Graph-RAG Architecture: Combines semantic search (RAG) with relationship graphs
  • Crash Dump Analysis: Parse stack traces and find problematic code instantly
  • Tree-sitter Parsing: Accurate C++ parsing that understands syntax
  • Incremental Indexing: Smart change detection and efficient updates
  • pgvector Storage: Fast vector similarity search
  • Web UI Dashboard: Configure directories and monitor indexing status
  • Docker + Podman: Works with both container runtimes

Quick Start

Prerequisites

  • Docker (20.x+) or Podman (4.x+)
  • 4GB+ RAM (8GB recommended for large codebases)
  • 5GB free disk space

1. Configure Environment

# Copy example config
cp env.example .env

# Edit .env to set your source code path
# Windows example: HOST_PATH=C:/Projects
# Linux example: HOST_PATH=/home/user/projects

2. Build and Start

Using Docker:

# Build images
docker-compose build

# Start services
docker-compose up -d

# View logs
docker-compose logs -f mcp-server

Using Podman:

# Build images
podman-compose build

# Start services
podman-compose up -d

# View logs
podman-compose logs -f mcp-server

3. Configure Directories

Open the web UI at http://localhost:8000 to:

  1. Browse your mounted directories
  2. Select folders to index
  3. Monitor indexing progress
  4. Test searches

4. Connect to Claude Desktop / VS Code

Add to your MCP client configuration:

Claude Desktop (%APPDATA%\Claude\claude_desktop_config.json on Windows):

{
  "mcpServers": {
    "cpp-codebase": {
      "url": "http://localhost:8000/mcp/v1"
    }
  }
}

VS Code (MCP extension settings):

{
  "mcp.servers": {
    "cpp-codebase": "http://localhost:8000/mcp/v1"
  }
}

Web UI

The dashboard at http://localhost:8000 provides:

Feature Description
Indexing Status View files indexed, entities found, progress
Directory Browser Navigate and select directories to index
Quick Search Test semantic search queries
Re-index Button Manually trigger re-indexing

Available MCP Tools

search_code

Semantic search across your codebase.

"Find database connection patterns"
"Show mutex locking implementations"

find_symbol

Precise symbol lookup with usages.

"Find ConnectionPool::acquire"
"Where is DatabaseManager defined?"

trace_dependencies

Graph traversal for dependencies.

"What does AuthManager depend on?"
"Show me everything that calls validateUser"

get_context

Comprehensive context for AI agents.

"Get context about the payment processing module"

analyze_debugging_context

Analyze crash dumps from Visual Studio.

Provide: file path, line number, exception info, call stack

find_code_location

Navigate to specific file and line.

"Show me database_connection.cpp line 95"

Configuration

Environment Variables

Variable Default Description
HOST_PATH ./example_code Path to mount as /host
MONITORED_PATHS (empty) Comma-separated paths to index
MCP_PORT 8000 Port for API and web UI
DB_NAME cpp_codebase PostgreSQL database name
DB_USER postgres Database user
DB_PASSWORD postgres Database password
EMBEDDING_MODEL all-MiniLM-L6-v2 Sentence transformer model

Volume Mounting

Mount your source code as read-only:

volumes:
  # Mount entire drive (Windows)
  - C:/:/host:ro
  
  # Mount specific directory (Linux)
  - /home/user/projects:/host:ro

Then use the web UI to select specific subdirectories.

Architecture

┌─────────────────────────────────────────────────────────────┐
│                    Container Network                         │
│  ┌─────────────────────┐    ┌─────────────────────────────┐ │
│  │  PostgreSQL 18      │    │  MCP Server                 │ │
│  │  + pgvector         │◄───│  + FastAPI                  │ │
│  │  (pg18-trixie)      │    │  + Web UI                   │ │
│  │  Port: 5432         │    │  Port: 8000                 │ │
│  └─────────────────────┘    └─────────────────────────────┘ │
└─────────────────────────────────────────────────────────────┘

Commands Reference

Docker

# Start
docker-compose up -d

# Stop
docker-compose down

# View logs
docker-compose logs -f mcp-server

# Rebuild after code changes
docker-compose build --no-cache

# Reset database (deletes all indexed data)
docker-compose down -v
docker-compose up -d

# Check status
docker-compose ps

# Enter container shell
docker-compose exec mcp-server bash

Podman

# Start
podman-compose up -d

# Stop
podman-compose down

# View logs
podman-compose logs -f mcp-server

# Rebuild
podman-compose build --no-cache

# Reset database
podman-compose down -v
podman-compose up -d

API Endpoints

# Check health
curl http://localhost:8000/api/status

# List MCP tools
curl http://localhost:8000/mcp/v1/tools

# Search code
curl -X POST http://localhost:8000/api/search \
  -H "Content-Type: application/json" \
  -d '{"query": "database connection"}'

# Get directories
curl http://localhost:8000/api/directories

# Browse directory
curl "http://localhost:8000/api/browse?path=/host"

Troubleshooting

Server won't start

# Check logs
docker-compose logs mcp-server

# Common issues:
# - PostgreSQL not ready: Wait 30 seconds
# - Port in use: Change MCP_PORT in .env
# - Out of memory: Increase Docker memory limit

No files indexed

# Verify mount
docker-compose exec mcp-server ls -la /host

# Check configured paths
curl http://localhost:8000/api/directories

Slow indexing

# Use faster embedding model
EMBEDDING_MODEL=sentence-transformers/all-MiniLM-L6-v2

# Check database size
docker-compose exec postgres psql -U postgres -d cpp_codebase \
  -c "SELECT pg_size_pretty(pg_database_size('cpp_codebase'));"

Permission denied on Linux

# Add user to docker group
sudo usermod -aG docker $USER

# Or use podman (rootless by default)
podman-compose up -d

Performance Tips

Codebase Size RAM First Index Time
10K LOC 4GB ~30 seconds
100K LOC 4GB ~5 minutes
1M LOC 8GB ~45 minutes
3M+ LOC 16GB ~2 hours

For Large Codebases

  1. Start with a single module to test
  2. Use the fast embedding model (default)
  3. Index only needed directories via web UI
  4. Consider SSD for database volume

Project Structure

cpp-graph-rag-mcp/
├── server.py              # Main MCP server (FastAPI)
├── parser.py              # Tree-sitter C++ parser
├── indexer.py             # Code indexer
├── crash_analyzer.py      # Crash dump analysis
├── vs_context_analyzer.py # VS debugging integration
├── config_manager.py      # Configuration persistence
├── requirements.txt       # Python dependencies
├── Dockerfile             # Container build
├── docker-compose.yml     # Multi-container setup
├── env.example            # Configuration template
├── static/                # Web UI files
│   ├── index.html
│   ├── styles.css
│   └── app.js
├── example_code/          # Sample C++ for testing
└── docs/                  # Documentation

Security Notes

  • Server runs on localhost only by default
  • Code is mounted read-only
  • Database password should be changed for production
  • No data leaves your machine (local embeddings)

License

MIT License - Free to use and modify.

Contributing

Key areas for improvement:

  • Template specialization handling
  • Macro expansion tracking
  • Multi-language support
  • Visual dependency graph viewer

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured