attest-mcp
Enables AI agents to scan payment endpoints for safety, returning a letter grade (A–F) and verdict before authorizing payments.
README
Attest MCP Server
Trust scanning for agent payments — right inside your AI agent.
attest-mcp is a Model Context Protocol server that lets any MCP-compatible agent (Claude, Cursor, Windsurf, VS Code, and more) scan an agent payment endpoint and get back a letter grade A–F with a clear safety verdict before authorizing a single cent.
It covers the agentic payment protocols in use today: x402, MPP, AP2, L402, and HTTP 402.
Powered by Attest. The scoring engine runs entirely on Attest's servers — this package is a thin client that relays requests, so installing it never exposes any proprietary grading logic.
Why
Autonomous agents are starting to pay for things on their own. A single malicious or misconfigured endpoint can drain funds through impersonation, a blocklisted payout wallet, bait-and-switch pricing, or a broken payment handshake. attest-mcp gives your agent a fast, deterministic safety check it can run as a step in its payment loop.
Tools
| Tool | What it does |
|---|---|
attest_scan |
Runs a full scan on a payment endpoint URL and returns a grade (A–F), a composite score (0–100), a verdict, danger flags, price, and a link to the full report. Call this before paying an unfamiliar endpoint. |
attest_grade |
Fast lookup of the most recent grade for a host that has already been scanned. Good for a quick pre-check. |
Requirements
- Node.js 18 or newer
- Any MCP-compatible client
No API key required.
Quick start
Run it directly with npx (no install needed):
npx attest-mcp
The server speaks MCP over stdio, so you normally don't run it by hand — you point your MCP client at it using one of the configs below.
Client setup
Claude Desktop
Edit claude_desktop_config.json (Settings → Developer → Edit Config):
{
"mcpServers": {
"attest": {
"command": "npx",
"args": ["-y", "attest-mcp"]
}
}
}
Cursor
Add to ~/.cursor/mcp.json (or Settings → MCP → Add new server):
{
"mcpServers": {
"attest": {
"command": "npx",
"args": ["-y", "attest-mcp"]
}
}
}
Windsurf
Add to ~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"attest": {
"command": "npx",
"args": ["-y", "attest-mcp"]
}
}
}
VS Code
Add to .vscode/mcp.json in your workspace:
{
"servers": {
"attest": {
"command": "npx",
"args": ["-y", "attest-mcp"]
}
}
}
Any other MCP client
Use the command npx -y attest-mcp with the stdio transport.
Remote server (no install)
Prefer not to install anything? Attest also runs a hosted MCP server you can connect to over Streamable HTTP:
https://attestagent.org/api/mcp
Point any remote-MCP-capable client at that URL.
Usage example
Once connected, just ask your agent in natural language:
"Before you pay, scan
https://api.example.com/paid-resourcewith Attest."
The agent calls attest_scan and gets back something like:
{
"host": "api.example.com",
"grade": "A",
"composite": 95,
"danger": false,
"verdict": "Valid endpoint, established host.",
"priceHuman": "0.01 USDC",
"reportUrl": "https://attestagent.org/r/abc-123"
}
Configuration
| Environment variable | Default | Description |
|---|---|---|
ATTEST_BASE_URL |
https://attestagent.org |
Override the Attest API base URL. Only needed for self-hosting or testing. |
Security & privacy
- No secrets, no accounts, no API keys. The server holds no credentials.
- No proprietary logic ships here. Grading happens server-side; this package only relays requests and formats responses.
- Inputs are validated before any request is made.
- Outbound only. The server makes HTTPS requests to the Attest API and speaks MCP over stdio — it opens no inbound ports.
- Scanned URLs and results are logged to Attest's public directory. Do not scan endpoints whose URL or response contains private or sensitive information. See attestagent.org/legal.
Links
- Website: attestagent.org
- Developer docs: attestagent.org/developers
- Methodology: attestagent.org/methodology
- X / Twitter: @Attestagent
Contributing
Issues and pull requests are welcome — see CONTRIBUTING.md.
License
MIT © Attest
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。