AgentRisk MCP
Official# AgentRisk MCP
Stop your AI agent from getting rekt. The pre-trade safety net it didn't know it needed — until it almost bought a honeypot. Live sell simulation across every DEX on Base (Uniswap V2/V3/V4, Aerodrome), catching traps a static code scan will never see.
## What it does
An MCP server that gives any MCP-compatible agent (Claude, Cursor, and others) a `check_token_risk` tool. Before your agent trades a token on Base, it can call this tool to get:
- Honeypot detection
- Deployer wallet freshness
- Brand impersonation checks
- On-chain LP-lock verification
- Live sell simulation across Uniswap V2/V3/V4 and Aerodrome
- A risk score, risk level, and clear should-trade decision
## Install
```bash
npm install -g agentrisk-mcp
```
## Usage with Claude Desktop
Add to your `claude_desktop_config.json`:
```json
{
"mcpServers": {
"agentrisk": {
"command": "npx",
"args": ["-y", "agentrisk-mcp"]
}
}
}
```
## Pricing
Paid per call via the x402 protocol, 0.15 USDC on Base, no API key, no subscription. Free public checks at [agentrisk.dev](https://agentrisk.dev) (3 free scans per wallet).
## Links
Website: https://agentrisk.dev
API repo: https://github.com/Neurobyteio/agentrisk
TDQS
Scored across 1 tool
With only a single tool, there is no possibility of ambiguity or overlap between tools. The tool's purpose is uniquely defined.
The tool name follows a clear verb_noun pattern ('check_token_risk'), and with only one tool there are no inconsistent naming conventions to flag.
A single tool feels thin for a server, even though the tool itself is comprehensive and well-scoped. It sits at the borderline of being too few, but the tool's depth partially compensates for the lack of additional tools.
For the stated purpose of token risk checking on Base, the tool covers a wide range of critical checks including honeypot detection, deployer history, impersonation, LP-lock verification, and live sell simulation. It provides a clear decision, leaving no obvious dead end within its domain.