Skip to main content
Glama
deenesjakoruzh
kvochkin-dev

MCP Gateway

by kvochkin-dev
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python

MCP Gateway

MCP Gateway — production-ready Model Context Protocol gateway for Russian LLMs: a unified interface to GigaChat and YandexGPT with automatic fallback, personal data anonymization per 152-FZ, and 86% savings on tokens.

Connect any MCP client (Hermes Agent, Claude Code, Cursor, n8n) — and get access to Russian models through a single endpoint. Stack: FastAPI, MCP SDK, GigaChat API, YandexGPT API.

✨ Features

Feature

Description

🔀 Multi-provider

GigaChat (primary) + YandexGPT (fallback) + Ollama (local)

🛡️ 152-FZ

Built-in detection and anonymization of personal data: phone, INN, passport, SNILS, address

💰 86% savings

GigaChat ~200₽/M tokens vs YandexGPT ~1,455₽/M tokens

🔄 Auto-fallback

Seamless provider switching on errors and rate-limit (429)

🐳 Docker Ready

Full Docker Compose for one-command deployment

💚 Health Checks

Monitoring endpoints and systemd service with auto-start

Related MCP server: Multi-Provider LLM MCP Server

🏗️ Architecture

flowchart TB
    classDef client fill:#eef2f7,stroke:#4a5568,stroke-width:1.5px
    classDef gateway fill:#edf2f7,stroke:#4a5568,stroke-width:1.5px
    classDef llm fill:#e2e8f0,stroke:#4a5568,stroke-width:1.5px

    subgraph Client["👤 Клиент"]
        Agent["🤖 MCP-клиент<br>Hermes Agent / Claude Code / n8n"]
    end
    class Client client

    subgraph Gateway["⚙️ MCP Gateway (FastAPI, port 8000)"]
        MCP["MCP Server<br>5 tools"]
        Anon["🛡️ Анонимизатор ПД<br>152-ФЗ"]
        Fallback["🔀 Fallback-оркестратор"]
        MCP --> Anon --> Fallback
    end
    class Gateway gateway

    subgraph Providers["🧠 LLM-провайдеры"]
        Giga["GigaChat<br>~200₽/M токенов"]
        Yandex["YandexGPT<br>~1 455₽/M токенов"]
        Ollama["Ollama<br>локально, 0₽"]
    end
    class Providers llm

    Agent --> MCP
    Fallback -->|"основной"| Giga
    Fallback -.->|"fallback"| Yandex
    Fallback -.->|"fallback"| Ollama

🚀 Quick Start

# Клонирование
git clone https://github.com/kvochkin-dev/mcp-gateway.git
cd mcp-gateway

# Настройка окружения
cp .env.example .env
# Впишите свои API-ключи в .env

# Запуск через Docker
docker-compose up -d

# Или напрямую
python -m venv venv && source venv/bin/activate
pip install -r requirements.txt
python app.py

📋 API Endpoints

Method

Path

Description

GET

/health

Service and provider status

GET

/docs

Swagger UI

POST

/mcp

MCP interface (tools)

Health Check

curl http://localhost:8000/health
{
  "status": "healthy",
  "gigachat": "connected",
  "yandexgpt": "connected",
  "anonymizer": "ready"
}

🔐 Configuration

All secrets are stored in .env (file excluded from git):

GIGACHAT_CLIENT_ID=***
GIGACHAT_CLIENT_SECRET=***
YANDEXGPT_API_KEY=***
YANDEXGPT_FOLDER_ID=***

🧪 Tests

pytest tests/ -v

13/13 tests passed — provider clients, fallback logic, personal data anonymizer.

📁 Project Structure

mcp-gateway/
├── app.py                  # FastAPI + MCP сервер
├── src/
│   ├── clients/            # GigaChat, YandexGPT клиенты
│   └── anonymizer/         # 152-ФЗ анонимизация ПД
├── scripts/
│   └── mcp-monitor.sh      # Мониторинг сервисов
├── docs/                   # Отчёты и документация
├── tests/                  # 13 тестов
└── docker-compose.yml      # Docker-развёртывание

📄 License

MIT — use freely.

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityStale
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • Model Runner

    joecheofficial
    A
    license
    Not graded
    quality
    D
    maintenance
    Provides a unified MCP interface for running completions, embeddings, image generation, and classification across OpenAI, Anthropic, Groq, and Mistral. Eliminates provider-specific boilerplate by standardizing API calls for text generation, vector embeddings, and classification tasks.
    MIT

  • Multi-Provider LLM MCP Server

    itachiuchihadev
    F
    license
    A
    quality
    D
    maintenance
    Unified MCP server for querying multiple LLM providers (Gemini, OpenAI, Anthropic, etc.) with advanced routing, cost optimization, and fallback resilience.
    2
    16 npm
    -

  • Ollama MCP Server

    claudiayoung104
    A
    license
    Not graded
    quality
    A
    maintenance
    Enables MCP-compatible applications to interact with local and cloud-hosted LLM models for model management, text generation, chat, embeddings, and web search/fetch.
    958 npm
    AGPL 3.0