Execute.run

Enables AI agents to manage Shell balances, transfer funds, and execute LLM requests via the Execute.run platform.

javascript

0

GitHub Stars

javascript

Language

4 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "execution-run-execution-run-mcp": {
      "command": "npx",
      "args": [
        "execution-run-mcp"
      ],
      "env": {
        "EXECUTE_RUN_API_KEY": "YOUR_API_KEY_PLACEHOLDER"
      }
    }
  }
}

This MCP server enables AI agents to interact with the Execute.run platform by managing Shell balances, transferring funds, and performing LLM requests through a straightforward, command-line driven interface.

How to use

You will run the MCP server locally and connect your MCP client (such as Claude Desktop or Kiro) to it. The server exposes a set of tools that let you query balances, view transactions, transfer Shells, sign challenges, and perform generated content requests with an LLM. To use it, start the MCP server with your API key and configure your MCP client to load this server as a named MCP source.

Key steps to use the server after installation: set your API key in the client configuration, start the MCP process, and then invoke the provided tools via your MCP client. The tools include whoami, get_balance, get_transactions, transfer, sign, and compute. Each tool performs a specific action within the Execute.run Shell economy and LLM execution flow, allowing you to manage assets and request AI model responses programmatically.

How to install

Choose your preferred installation method from the options below. All methods require Node.js tooling to be available on your system as described.

# Recommended: run via npx without global install
npx @execution-run/mcp-server

Alternatively, install globally with npm and then start the server directly from the command line.

npm install -g @execution-run/mcp-server
execution-run-mcp

If you prefer building from source, clone the repository, install dependencies, and build the project. Then you can run the built server locally.

git clone https://github.com/execution-run/execution-run-mcp.git
cd execution-run-mcp
bun install
bun run build

When you run the server, you must provide your Execute.run API key. The API key can be supplied through environment configuration in your MCP client or by running the server in a context that exports the key.

Configuration and usage notes

Environment variables you need to set when running the MCP server: EXECUTE_RUN_API_KEY (required) with your key from execution.run/api-keys. Optionally, you can set EXECUTE_RUN_API_URL to override the API base URL; the default is https://execution.run/v1.

Client integrations shown in examples configure the server by referencing the MCP executable and passing the API key via environment variables. For Claude Desktop and Kiro, you typically specify the MCP server as a local node that runs the MCP client binary with the appropriate environment variables.

Standalone usage example shows how to export the API key in the environment and start the MCP server directly.

Tools and endpoints overview

The server exposes a set of tools that you can invoke through the MCP interface: whoami, get_balance, get_transactions, transfer, sign, and compute. Each tool has a distinct purpose, from retrieving identity and balances to performing transfers and executing LLM requests.

Available tools

whoami

Retrieve your card and wallet identity information including IDs, address, balance, and status.

get_balance

Return the current Shell balance and the maximum allowed balance for your wallet.

get_transactions

Fetch the transaction history for your wallet with optional limit control.

transfer

Transfer Shells to another wallet or card address, including amount and purpose.

sign

Sign a challenge with your card's Ed25519 keypair and return the signature along with identifiers.

compute

Execute an LLM request by burning Shells, specifying model, messages, temperature, and token limit.