near MCP Server

Use This MCP server To

Query NEAR blockchain account data in real time Execute blockchain transactions via LLM commands Integrate NEAR blockchain context into AI workflows Automate decentralized app interactions with AI agents Monitor NEAR blockchain events for alerts or triggers Fetch smart contract states for analysis or reporting

README

NEAR MCP

This project is a Model Context Protocol (MCP) compatible server for interacting with the NEAR blockchain. This tool provides a way for LLMs and AI agents to securely access and interact with NEAR accounts and blockchain functionality.

Quickstart

Here is how to get started with the near-mcp server quickly with the claude code cli

npm install -g @anthropic-ai/claude-code
claude mcp add near-mcp npx @nearai/near-mcp@latest run
claude

Or deploy the MCP server remotely on Phala Cloud, check the instructions here

Installing

near-mcp is meant to be used is with an MCP compatible client. Learn more in the MCP docs

Adding to the claude code cli:

claude mcp add near-mcp npx @nearai/near-mcp@latest run

Adding to claude desktop via JSON config:

{
  "mcpServers": {
    "near-mcp": {
      "command": "npx",
      "args": ["-y", "@nearai/near-mcp@latest", "run"],
      "env": {}
    }
  }
}

Adding to goose

┌   goose-configure
│
◇  What would you like to configure?
│  Add Extension
│
◇  What type of extension would you like to add?
│  Command-line Extension
│
◇  What would you like to call this extension?
│  near-mcp
│
◇  What command should be run?
│  npx @nearai/near-mcp@latest run
│
◇  Please set the timeout for this tool (in secs):
│  60
│
◇  Would you like to add environment variables?
│  No
│
└  Added near-mcp extension

Or you can install it globally and use it directly.

# Install globally
npm install -g @nearai/near-mcp@latest

# Or use directly with npx
npx @nearai/near-mcp@latest run

Available Tools

see TOOLS.md for a list of available tools and arguments.

Integration with AI Models

This tool is designed to be used with AI models that support the Model Context Protocol. It enables AI assistants to:

Manage NEAR accounts on behalf of users
Check account balances and status
Sign and send transactions
Create new accounts and manage access keys
Inspect and execution smart contracts

Security Considerations

This MCP is meant to be run locally. Account private keys are stored in a local unencrypted keystore where the MCP server is running.
The underlying models should not have access to see the private keys of the accounts they are interacting with with one exception. The import_account tool allows the model to import an account from a private key. This requires the user to provide the private key to the model.

Contributing

We welcome contributions to the NEAR MCP server! Please see the CONTRIBUTING.md file for more information.

Reporting Issues

If you find a bug or have a feature request, please open an issue on the GitHub repository.

near-mcp FAQ

How do I install near-mcp?

Install via npm with 'npm install -g @nearai/near-mcp' or run it using 'npx @nearai/near-mcp@latest run'.

Can near-mcp be deployed remotely?

Yes, near-mcp can be deployed on cloud platforms like Phala Cloud for remote access.

What blockchain does near-mcp support?

near-mcp supports the NEAR blockchain, enabling interaction with NEAR accounts and smart contracts.

How does near-mcp ensure secure blockchain interactions?

near-mcp uses MCP standards to securely expose blockchain data and functions to LLMs with scoped access.

Which LLMs can work with near-mcp?

near-mcp is compatible with LLMs like OpenAI GPT-4, Anthropic Claude, and Google Gemini via MCP protocol.

Is near-mcp suitable for production use?

Yes, near-mcp is designed for both development and production environments with secure, scalable blockchain access.

How do I add near-mcp to my MCP client?

Use the MCP client CLI command 'claude mcp add near-mcp npx @nearai/near-mcp@latest run' to integrate it.

Does near-mcp support smart contract interaction?

Yes, it allows querying and invoking NEAR smart contracts through structured MCP calls.