SafeAgent Shield

The firewall between AI agents and the blockchain. Free during beta — no API key, no payment, no limits.

Every crypto action your agent takes should go through Shield first. It checks safety, simulates the transaction, and returns GO or BLOCK.

Without Shield: Agent → Blockchain → might lose everything
With Shield:    Agent → Shield → Blockchain → guaranteed safe

One Tool. Complete Protection.

# MCP — one call protects everything
shield(action="buy", token="0x...", chain="base")

# Returns:
# 🛡️ SAFEAGENT SHIELD — BUY
# Target: DEGEN (DEGEN)
# Safety: 90/100 — LIKELY SAFE
# ✅ Sell verified — tax: 0%
# DECISION: ✅ GO — Score 90/100. Safe.

Actions: buy | sell | approve | interact | check

What Shield Does

Check	Method	Time
Honeypot detection	Real DEX swap simulation (not code guessing)	~2s
27 scam patterns	Source code analysis	<1s
Approval safety	Contract verification + phishing detection	<1s
Tax calculation	Exact buy/sell tax via swap sim	~2s
Liquidity check	DEX pair verification	<1s

Key difference: Other tools predict honeypots from code patterns. Shield tests them with real DEX swaps. Not guessing — proving.

Connect

MCP Streamable HTTP: POST https://cryptogenesis.duckdns.org/mcp
MCP SSE:             GET  https://cryptogenesis.duckdns.org/mcp/sse
REST API:            GET  https://cryptogenesis.duckdns.org/token/scan
Smithery:            npx @smithery/cli install @safeagent/token-safety

23 Tools

Shield (primary): shield — GO/BLOCK decision for any crypto action

Safety: test_honeypot · check_token_safety · check_before_buy · check_approval_safety · simulate_swap · check_wallet_risk

Market: get_token_price · get_trending_tokens · get_gas_prices · get_new_tokens · get_defi_yields · get_market_overview · get_defi_tvl

Agent tools: get_portfolio · get_chain_info · get_eth_balance · resolve_ens · search_token · create_agent_token

System: ping

On-Chain (ERC-7913)

Chain	Oracle	Router	Factory
Base	0x37b9...8e	0xb200...FA	—
Optimism	0x3B8A...47	—	0x9B4A...84

For Smart Contracts

import {SafeGuard} from "@safeagent/guard/contracts/SafeGuard.sol";

contract MyDEX {
    using SafeGuard for address;
    function swap(address tokenOut, uint256 amt) external {
        tokenOut.requireSafe(); // 1 line. Done.
    }
}

SafeGuard Library →

Standard

ERC-7913 Token Safety Score — the open standard for token safety on EVM chains.

License

MIT