Chef

An AI-powered competitive programming assistant implementing the Model Context Protocol (MCP). This server allows AI agents (like Claude Desktop or Cursor) to participate in CodeChef contests: fetching problems, generating/testing solutions in a secure sandbox, managing retry states, submitting answers, and tracking contest progress.

---

🚀 Features

• Contest Management: Open contests, retrieve problem lists, and track solved vs. remaining problems.
• Browser Automation: Playwright-based logins, problem-statement scraping, submission uploads, and verdict polling.
• Secure Sandbox: Docker-isolated compilation and execution (supports C++, Python, Java, Go, Rust) with no internet access, limited CPU/memory, and execution timeouts.
• Validation & Retry: Custom edge-case generator, confidence evaluator, and self-repair engine to analyze verdicts and repair failing solutions iteratively.

---

📂 Project Structure

MCP_Chef/
│
├── app/
│   ├── browser/       # Playwright-based browser automation (login, submit, poll)
│   ├── models/        # SQLAlchemy database schema (SQLite)
│   ├── retry_engine/  # Diagnosis & self-repair pipeline
│   ├── sandbox/       # Secure Docker runner configuration & language setup
│   ├── solver/        # Sequential contest solver state (Q1 → Q5)
│   ├── tools/         # Exposed MCP tool decorators
│   ├── utils/         # Structured logger and cache layer
│   ├── main.py        # Streamable HTTP ASGI app
│   └── mcp_server.py  # FastMCP server instantiation
│
├── tests/             # Automated test suite
├── Dockerfile         # Multi-stage container definition
├── docker-compose.yml # Compose configuration (App, Redis)
├── run_server.py      # Unified server CLI entrypoint (STDIO & HTTP)
└── .env               # Local configuration file (not tracked in Git)

---

🛠️ Local Setup

Prerequisites

• Python 3.10+
• Docker (Required for sandbox execution)
• Node.js (Optional, for localtunnel testing)

Step 1: Install Dependencies

Create a virtual environment and install dependencies:

python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
playwright install chromium

Step 2: Configure Environment

Copy the example environment file and add your CodeChef credentials:

cp .env.example .env

Open .env and set your CodeChef username and password.

Step 3: Run Tests

Verify your local installation:

python -m pytest tests/ -v

---

💻 Running the Server

1. STDIO Transport (For Claude Desktop)

To run the server locally over Standard Input/Output:

python run_server.py

Add the configuration to your Claude Desktop config (e.g., ~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "codechef": {
      "command": "/Users/ayushkumarsingh/Downloads/MCP_Chef/venv/bin/python",
      "args": ["/Users/ayushkumarsingh/Downloads/MCP_Chef/run_server.py"]
    }
  }
}

2. HTTP Transport (For Cursor/Web Clients)

To run the server over HTTP/SSE:

python run_server.py --http --port 8000

Then configure your client to connect to: http://localhost:8000/mcp

---

☁️ Deployment

We have configured deployment setups for both container and VM platforms:

• Render (Free Tier): For a quick deployment that does not require a card, see the Render Deployment Guide.
• Fly.io (VM with Sandbox): For running with the secure Docker sandbox enabled, see the Fly.io Deployment Guide.