Block MCP Server

⛓️ Block MCP Server

A Model Context Protocol (MCP) server for Blockchain interactions. This server enables AI agents to interact with multiple blockchain networks, check financial data, and send real-time notifications to Telegram.

🛠️ Tools Included

1. 💰 Wallet Balance Checker: Retrieves native token balances for any wallet address across Ethereum, Polygon, Arbitrum, Optimism, and Base simultaneously using public RPC nodes.
2. 📜 Frax Transactions: Fetches the 5 latest transactions on the FRAX network, including both native transfers and ERC-20 token events.
3. ⛽ Gas Price Estimator: Provides real-time gas fee estimates (in Gwei) across multiple blockchain networks to help determine transaction costs.
4. 💱 Crypto Converter: Converts cryptocurrency and fiat values using real-time market rates (e.g., BTC to ETH or USDC to USD).
5. 🤖 Telegram Bot: Programmatically posts updates, messages, or automated notifications to a specific Telegram chat or channel.

⚙️ Setup Instructions

Prerequisites

• Python 3.11+
• uv

1. Environment Variables

Create a .env file in the root directory and populate it with your credentials:

# Telegram Configuration
TELEGRAM_BOT_TOKEN=your_bot_token_from_botfather
TELEGRAM_CHAT_ID=your_numeric_chat_id

# Blockchain API Keys
FRAXSCAN_API_KEY=your_fraxscan_api_key

# Optional: CoinGecko API Key
COINGECKO_API_KEY=your_api_key

2. Installation

1. Clone the repository:

   git clone <YOUR_REPO_URL_HERE>
   cd <YOUR_REPO_NAME>
   

2. Install dependencies: This project uses uv for fast dependency management.

   make install
   

🚀 How to Run

The easiest way to test the tools is using the built-in Makefile command, which launches the MCP Inspector.

make dev

Alternatively: uv run fastmcp dev main.py

📖 Usage & Examples

Below are example inputs for testing the tools in the MCP Inspector.

1. Wallet Balance Checker Tool (`wallet_check`)

Queries multiple public RPC endpoints to retrieve native balances across various chains.

• Input:

  { "address": "0x..." }
  

• Response:

  {
    "Ethereum": "1.2450 ETH",
    "Arbitrum": "0.0000 ETH",
    "Polygon": "150.3200 MATIC",
    "Optimism": "0.0500 ETH",
    "Base": "0.1200 ETH"
  }
  

2. Frax Transaction History (`get_frax_transactions`)

Merges native and token transfer history from the Fraxscan API, sorted by timestamp.

• Input:

  { "address": "0x4200000000000000000000000000000000000015" }
  

• Response:

  [
    {
      "time": "2026-01-30 18:00:00",
      "amount": "0.5000 frxETH",
      "token": "frxETH",
      "type": "Native",
      "hash": "0x..."
    }
  ]
  

3. Gas Price Estimator (`get_gas_prices`)

Monitors real-time gas prices (in Gwei) to estimate transaction costs.

• Input:

  {}
  

• Response:

  {
    "Ethereum": "🔴 45.2 Gwei",
    "Base": "🟢 0.01 Gwei"
  }
  

4. Crypto Converter (`convert_crypto`)

Converts time between timezones.

• Input:

  {
    "amount": 1,
    "from_coin": "bitcoin",
    "to_coin": "ripple"
  }
  

• Response:

  {
    "Response": "1 BITCOIN = 1235999999.0 XRP (Rate: 95000)"
  }
  

5. Telegram Bot (`send_telegram_message`)

This tool allows the AI to send notifications to your phone via Telegram.

🔄 How the Flow Works (Sending Messages)

1. User Trigger: You give Claude a condition or a direct command (e.g., "If ETH is below $2000, alert me").
2. Decision: Claude evaluates the condition. If true, it decides to call send_telegram_message.
3. Execution: The MCP server receives the text payload and makes an HTTP POST request to the Telegram Bot API.
4. Confirmation: The Telegram API confirms the delivery, and the MCP server reports back to Claude that the task is done.

📝 Implementation Details

• Input:

{
  "message": "The deployment was successful"
}

• Response:

"Message sent successfully!"

🤖 Agent Mode (Auto-Pilot)

This mode enables the AI to automatically check for new messages and respond to them in a continuous loop, while also performing other operations in between.

Prompt to Initiate:

"I want you to run in Auto-Pilot Mode. Here is your loop:
1. **Check:** Run `telegram_read_messages`.
2. **Act:** If there is a NEW message (one you haven't answered yet), interpret it and use `telegram_reply` to send the answer.
3. **Wait:** If there are no new messages, carry out any pending commands then run `wait_for_seconds(4)`.
4. **Repeat:** Go back to Step 1.
Do not ask me for permission between steps. Just keep running this loop indefinitely."

How the Loop Works:

1. Check: The AI polls the Telegram API for incoming messages.
2. Process: It reads the content of any new user messages (e.g