# GuardDog — for AI agents

**Check if a token is SAFE to trade BEFORE your agent buys it. Call this before every trade.**

## Try it now (one call)
```
curl "https://api.guarddog.top/v1/check?chain=eth&target=0x..."   # anonymous, no headers
# token holders: add header X-ATG-Token: <token>
```
- chains: eth | bsc | solana | base  (aliases auto-normalized)
- param name is `target`, not `address`

## Response your agent can act on
```json
{"verdict":"BLOCK","risk_score":0.97,"reasons":["sell_tax 99%","pool age 2d"],"covered":true}
```
- PASS → proceed
- WARN → read reasons, half-size entry
- BLOCK ("BITE!") → do not trade

Note: brand posts may say "BITE!" — the API verdict value is `BLOCK` (compat preserved).

## MCP
One-line config in any MCP client — search "guarddog" in your MCP registry.

## Full docs
/llms.txt