# GuardDog

> Check if a token is safe before trading — one call, PASS/WARN/BITE.
> The on-chain guard dog for AI agents. 链上看门狗.

GuardDog (aka AgentTradeGuard) is an on-chain token safety API for AI agents.
It checks ETH / BSC / Solana / Base tokens against 4 data sources (GoPlus, Etherscan,
DEXScreener, honeypot.is) and returns a single actionable verdict.

## Verdicts
- PASS: safe to trade
- WARN: suspicious — read reasons, half-size entry recommended
- BLOCK (a.k.a. "BITE!"): honeypot/rug pattern — do not trade

## Quick start
```
curl "https://api.guarddog.top/v1/check?chain=eth&target=0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2" -H "anonymous, tokenless"
```
NOTE: the address parameter is named `target` (not `address`).
Chain values: eth | bsc | solana | base (aliases like "sol"/"ethereum" are auto-normalized).

## Key facts
- P95 latency: 23ms (hot path)
- Validated on ACM SolRPDS dataset: 116,304 labeled rug pulls, 100% data coverage
- MCP server available (search "guarddog" in MCP registries)
- Free during 90-day beta: 100 checks/day, no signup

## Pages
- https://guarddog.top/ — this site
- https://guarddog.top/AGENTS.md — agent-readable entry page
- https://x.com/OnchainGuardDog — daily threat feed & intercept reports