# @agentradar/mcp

[![npm version](https://img.shields.io/npm/v/@agentradar/mcp.svg)](https://www.npmjs.com/package/@agentradar/mcp)
[![Build](https://github.com/Bichev/agentradar-mcp/actions/workflows/build.yml/badge.svg)](https://github.com/Bichev/agentradar-mcp/actions/workflows/build.yml)
[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE)
[![MCP Registry](https://img.shields.io/badge/MCP%20Registry-io.github.Bichev%2Fagentradar-orange)](https://registry.modelcontextprotocol.io/v0.1/servers?search=io.github.Bichev/agentradar)

MCP Server for **AgentRadar** — the on-chain trust oracle for the agent economy. Provides 18 tools for verifying, scoring, and attesting AI agent trustworthiness directly from Claude, Cursor, or any MCP-compatible client.

- **Live API:** https://api.vvpro.ai
- **Web:** https://vvpro.ai
- **MCP Registry:** [`io.github.Bichev/agentradar`](https://registry.modelcontextprotocol.io/v0.1/servers?search=io.github.Bichev/agentradar)

This package is a thin client wrapper around the public AgentRadar API. The full source is auditable here — anything `npx @agentradar/mcp` runs is in this repo, MIT-licensed.

## Tools

| Tool | Description |
|------|-------------|
| `verify_agent` | Full 6-signal trust verification |
| `compare_agents` | Side-by-side comparison of two agents |
| `batch_verify` | Verify up to 10 agents at once |
| `get_score` | Quick cached trust score lookup |
| `get_badge_url` | Generate embeddable trust badge URL |
| `attest_agent` | Score + write on-chain EAS attestation |
| `get_attestations` | List all attestations |
| `get_attestation_by_tx` | Look up attestation by tx hash |
| `lookup_identity` | On-chain identity (ERC-8004) |
| `lookup_reputation` | Reputation signals |
| `resolve_agent_uri` | Resolve agent URI/metadata |
| `check_scam` | Check scam database |
| `add_scam_wallet` | Report malicious wallet |
| `explain_score` | Human-readable score explanation |
| `get_health` | API health check |
| `get_stats` | Dashboard statistics |
| `list_signals` | List scoring signals + weights |
| `get_pricing` | API pricing info |

## Installation

### Cursor

Add to your project's `.cursor/mcp.json`:

```json
{
  "mcpServers": {
    "agentradar": {
      "command": "npx",
      "args": ["-y", "@agentradar/mcp"],
      "env": {
        "AGENTRADAR_OPERATOR_KEY": "your-key-here"
      }
    }
  }
}
```

### Claude Desktop

Add to `~/Library/Application Support/Claude/claude_desktop_config.json` (macOS) or `%APPDATA%\Claude\claude_desktop_config.json` (Windows):

```json
{
  "mcpServers": {
    "agentradar": {
      "command": "npx",
      "args": ["-y", "@agentradar/mcp"],
      "env": {
        "AGENTRADAR_OPERATOR_KEY": "your-key-here"
      }
    }
  }
}
```

### npx (direct)

```bash
npx @agentradar/mcp
```

### Local development

```bash
git clone https://github.com/Bichev/agentradar-mcp.git
cd agentradar-mcp
npm install
npm run build
node dist/index.js
```

## Environment Variables

| Variable | Required | Default | Description |
|----------|----------|---------|-------------|
| `AGENTRADAR_API_URL` | No | `https://api.vvpro.ai` | API base URL |
| `AGENTRADAR_OPERATOR_KEY` | No | — | Operator key for admin endpoints and bypassing x402 payment |

## Usage Examples

Once connected, you can ask your AI assistant:

- "Verify this agent: 0x742d35Cc6634C0532925a3b844Bc9e7595f2bD28"
- "Is 0xdead...beef a known scam wallet?"
- "Compare these two agents and tell me which is more trustworthy"
- "Explain why this agent scored 45/100"
- "Write an on-chain attestation for this verified agent"

## Pricing

- **verify_agent**: $0.005 per call (x402) or free with operator key
- **attest_agent**: $5.00 per attestation (x402) or free with operator key
- **get_score**: Free
- **Admin endpoints**: Operator key required

## License

MIT