About AgentPay
AgentPay is an MCP server for agents that need to access paid APIs using USDC on Base. It combines Xiaohongshu note data and wholesale pricing tools with automatic x402 payment handling.
Key features
- 01Xiaohongshu note details and user posts
- 021688 and Yiwu wholesale price queries
- 03Automatic x402 payment signing and retry
- 04Payment status, wallet, and spending limits
Use cases
- 01Retrieve details for a Xiaohongshu note
- 02List a creator’s recent posts
- 03Check factory-direct wholesale prices
- 04Inspect wallet settings and spending limits
MCP support
Exposes paid API tools to MCP clients through remote Streamable HTTP or local stdio. The server handles x402 payment signing and retries before returning JSON results.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://agentpay-xhs-production.up.railway.app/mcpclaude mcp add --transport http yyxqqq777-agentpay https://agentpay-xhs-production.up.railway.app/mcp{"mcpServers":{"yyxqqq777-agentpay":{"type":"http","url":"https://agentpay-xhs-production.up.railway.app/mcp"}}}Open in Cursor{"servers":{"yyxqqq777-agentpay":{"type":"http","url":"https://agentpay-xhs-production.up.railway.app/mcp"}}}Open in VS Code{"mcpServers":{"yyxqqq777-agentpay":{"type":"http","url":"https://agentpay-xhs-production.up.railway.app/mcp"}}}{"mcpServers":{"yyxqqq777-agentpay":{"serverUrl":"https://agentpay-xhs-production.up.railway.app/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEAgentPay MCP Server
Agent-to-Agent micropayments via x402 + MCP — let AI agents autonomously pay for API access with USDC on Base.
Install
Option A — Remote MCP(推荐,无需本地安装)
在 Cursor / Claude / 任意支持 Streamable HTTP 的客户端添加:
https://agentpay-xhs-production.up.railway.app/mcp
官方 Registry 条目:io.github.yyxqqq777/agentpay
Option B — 本地 stdio(uvx / PyPI)
{
"mcpServers": {
"agentpay": {
"command": "uvx",
"args": ["mcp-server-agentpay"],
"env": {
"AGENT_PRIVATE_KEY": "0xYourAgentPrivateKey"
}
}
}
}
只需填付款钱包私钥(需有 USDC)。网关地址已内置,一般不用改。
或永久安装:
pip install mcp-server-agentpay
# or
uv tool install mcp-server-agentpay
Tools
| Tool | Description | Cost |
|---|---|---|
xhs_get_note_detail | Xiaohongshu note detail | 0.01 USDC/call |
xhs_get_user_notes | Xiaohongshu user posted notes list | 0.01 USDC/call |
china_wholesale_pricing_query | Factory-direct wholesale pricing (1688/Yiwu) | 0.01 USDC/call |
agentpay_payment_status | View payment config, wallet, and spending limits | Free |
Example prompts
- "帮我看看这条小红书笔记写了什么:https://www.xiaohongshu.com/explore/..."
- "拉一下这个博主最近发的笔记,user_id=..."
- "Search wholesale prices for wireless earbuds under $3"
- "What's my AgentPay wallet and spending limit?"
Environment Variables
| Variable | Required | Description |
|---|---|---|
AGENT_PRIVATE_KEY | Yes(付费工具) | 付款钱包私钥 |
MAX_SPEND_PER_CALL | No | 单次上限(默认 0.10) |
GATEWAY_BASE_URL | No | 网关地址(默认已指向公网 Gateway) |
How it works
MCP Client → mcp-server-agentpay → x402 Gateway → Paid API
(auto 402 → sign → retry) (verify USDC)
- Tool call hits a paid API endpoint
- Gateway returns
402 Payment Requiredwith USDC payment details - MCP server signs payment and retries automatically
- JSON data flows back to the LLM
Self-host the Gateway
If you operate the paid API, deploy the gateway separately:
pip install "mcp-server-agentpay[gateway]"
# .env: PAYMENT_MODE=production + upstream keys + your receive wallet
agentpay-gateway
Real USDC settlement (x402)
| Mode | Behavior |
|---|---|
PAYMENT_MODE=demo | Local fake pay (dev only) |
PAYMENT_MODE=production | Official Facilitator verify + settle on-chain |
Testnet (Base Sepolia) — use https://x402.org/facilitator, network eip155:84532.
# Fund a test agent with Sepolia USDC, then:
PYTHONPATH=src python scripts/test_x402_settlement.py
Mainnet — set PAYMENT_NETWORK=eip155:8453 and a mainnet facilitator
(https://api.cdp.coinbase.com/platform/v2/x402 or PayAI). USDC goes to WALLET_RECIPIENT_ADDRESS.
See .env.example.
Publish to MCP Registry
This package is designed for the official MCP Registry.
# 1. Publish to PyPI
pip install build twine
python -m build && twine upload dist/*
# 2. Publish to MCP Registry (packages + remotes)
./scripts/publish_all.sh
Or create a GitHub Release (v0.1.0) — Actions uses Trusted Publishing to PyPI then updates the Registry.
Before publishing, update server.json and README mcp-name with your GitHub namespace (io.github.<username>/agentpay).
Development
git clone https://github.com/yyxqqq777/mcp-server-agentpay
cd mcp-server-agentpay
python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"
pytest tests/ -v
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.