门罗币的MCP服务器。给任何AI代理一个私人钱包。
概览
这是人工智能代理与门罗网络之间的桥梁。它允许任何人工智能代理——OpenClaw、Agent Zero、Claude或任何兼容MCP的框架——通过简单的工具界面检查余额、接收付款并发送XMR。 可以这样想：你的AI代理拥有自己的门罗币钱包。 MCP（模型上下文协议）是由Anthropic创建的标准，允许AI助手使用外部工具。AI不仅仅是生成文本，而是调用真实功能——读取文件、查询数据库，或者在这里与Monero钱包交互。 如果你曾经希望你的AI代理能够自主接收和发送XMR——无论是OpenClaw代理运行服务、Agent Zero实例管理自己的预算，还是仅仅是Claude检查你的余额——这就是它的作用。 因为人工智能代理比人类更需要隐私。 当人类进行交易时，自然会产生噪音——你在随机时间买咖啡，不定期付房租，你的模式混乱。人工智能智能体则不同。
概述（AI 翻译）
全球首个门罗币主控服务器。给任何AI代理一个私人钱包。
这是什么？
这是人工智能代理与门罗网络之间的桥梁。它允许任何人工智能代理—— OpenClaw、Agent Zero、Claude 或任何兼容 MCP 的框架——校验平衡， 通过简单的工具界面接收付款并发送XMR。
可以这样想：你的AI代理拥有自己的门罗币钱包。
等等，MCP是什么？
MCP（模型上下文协议）是由Anthropic创建的标准，允许AI助手使用外部工具。AI不仅仅是生成文本，而是调用真实功能——读取文件、查询数据库，或者在这里与Monero钱包交互。
如果你曾希望你的AI代理能够自主接收和发送XMR—— 无论是运行服务的OpenClaw代理，还是Agent Zero实例 管理自己的预算，或者只是Claude帮你检查余额——这就是它的作用。
为什么是门罗币？为什么不是以太坊或比特币？
因为人工智能代理比人类更需要隐私。
当人类进行交易时，自然会产生噪音——你在随机时间买咖啡，不定期付房租，你的模式很混乱。人工智能智能体则不同。每笔交易都遵循逻辑。
本页由 AI 翻译，技术细节请以原文为准。
README
以下为原文 README 节选，完整内容请查看原文。
🔒 Monero MCP Server
The world’s first MCP server for Monero. Give any AI agent a private wallet.
What is this?
This is a bridge between AI agents and the Monero network. It lets any AI agent – OpenClaw, Agent Zero, Claude, or any MCP-compatible framework – check balances, receive payments, and send XMR through a simple tool interface.
Think of it as: your AI agent gets its own Monero wallet.
Wait, what’s MCP?
MCP (Model Context Protocol) is a standard created by Anthropic that lets AI assistants use external tools. Instead of the AI just generating text, it can call real functions – read files, query databases, or in this case, interact with a Monero wallet.
If you’ve ever wanted your AI agent to autonomously receive and send XMR – whether it’s an OpenClaw agent running a service, an Agent Zero instance managing its own budget, or just Claude checking your balance – that’s what this does.
Why Monero? Why not Ethereum or Bitcoin?
Because AI agents need privacy more than humans do.
When a human makes transactions, there’s natural noise – you buy coffee at random times, pay rent irregularly, your patterns are chaotic. An AI agent is different. Every transaction follows logic. On a transparent blockchain, anyone can watch an agent’s on-chain behavior and reverse-engineer its strategy, its triggers, and how to exploit it.
Monero’s privacy-by-default design means an agent’s transaction history, balance, and counterparties are hidden. That’s not a nice-to-have – it’s a security requirement for any autonomous agent handling real value.
Privacy is not a feature for AI agents. It’s infrastructure.
Features
12 tools covering everything an AI agent needs:
| Category | Tools |
|---|---|
| Balance & Info | get_balance, get_address, get_height, get_version |
| Transactions | get_transfers, get_transfer_by_txid |
| Receiving | create_address, make_integrated_address |
| Sending | transfer, sweep_all, confirm_transfer |
| Validation | validate_address |
| Optional (opt-in) | get_exchange_rate (XMR/USD, XMR/EUR; set MONERO_ENABLE_PRICE_FEED=true) |
Read-only by default. Sending XMR requires explicit opt-in via environment variable.
Security: Built for a hostile world
This server assumes the AI agent will encounter prompt injection attacks – malicious text designed to trick the AI into sending funds to an attacker. The AI is not our security boundary. The MCP server is.
Five defense layers, all configurable:
🛡️ Address Allowlist – Lock transfers to known addresses only. Prompt injection can’t send to unknown wallets.
🛡️ Two-Step Confirmation – Transfers return a preview + one-time token. Must call confirm_transfer to execute. Attackers can’t predict the token.
🛡️ Rate Limiting – Cooldown between transfers + daily XMR limit. Limits damage even if everything else fails.
安装
npx -y monero-mcp@latest配置
{
"mcpServers": {
"monero": {
"command": "npx",
"args": ["-y", "monero-mcp@latest"],
"env": {
"MONERO_RPC_HOST": "127.0.0.1",
"MONERO_RPC_PORT": "18082",
"MONERO_NETWORK": "mainnet",
"MONERO_ALLOW_TRANSFERS": "false"
}
}
}
}