# Steve Guard — Pre-Trade Safety Layer for Autonomous Agents

Robinhood-chain (4663) & Solana token SAFETY + dollar-weighted NET-USD
accumulation FOOTPRINT. Call this BEFORE your agent executes a swap: is the
token a honeypot/rug, and are real dollars/repeat wallets ACCUMULATING or is it
DISTRIBUTION? Dollar-weighted — avoids the trade-count trap that fools naive
rug-checkers and DexScreener-only agents.

## Why an execution agent should call this first
Autonomous swap agents (Vex-style, coding-agent + wallet, etc.) are strong at
EXECUTION but usually rely on basic market data. We are the specialized safety/
intelligence layer: honeypot via REAL buy+sell simulation + net-USD footprint.
Example: a token can look fine on price/volume while net-USD flow is already
distribution (money leaving) — we surface that; count-based tools do not.

## Endpoints (free tier, no key)
- GET https://api.stevesignals.com/score_rh?token=0x...   -> Robinhood-chain safety + net-USD footprint
- GET https://api.stevesignals.com/score_sol?token=<mint> -> Solana safety + net-USD footprint
- GET https://api.stevesignals.com/v1/x402/robinhood/honeypot/0x...  -> x402-paid honeypot ($0.02, USDG/USDC)
- MCP (Streamable HTTP): https://api.stevesignals.com/mcp  (tool: check_robinhood)
- Discovery: https://api.stevesignals.com/.well-known/x402

## Add to your agent (one step)
MCP server config (Claude Code / Cursor / any MCP client):
  {"mcpServers": {"steve-guard": {"url": "https://api.stevesignals.com/mcp"}}}
Then your agent can call check_robinhood(token) before any Robinhood-chain trade.
Or just curl the REST endpoint above — no key, no signup.

## Honesty
This is a risk/behaviour SCREEN, not financial advice and not a profit predictor.
We sell detection quality, not trade calls. No private keys held server-side.

## Example
curl "https://api.stevesignals.com/score_rh?token=0xYOUR_TOKEN"