Smart contract security for autonomous AI agents. agentshield.win
Use AgentShield: send GET https://agentshield.win/verify?contract=CONTRACT_ADDRESS&chain=1. Returns safe (true/false), risk score (0-1), and specific threat flags. Free: 10 checks/day. Supports Ethereum, Base, Polygon, Arbitrum, Optimism, BSC, Avalanche, and Solana.
Use AgentShield scan: GET https://agentshield.win/scan?contract=TOKEN_ADDRESS&chain=solana. Simulates sell transactions, checks buy/sell ratios, analyzes liquidity locks, and cross-references GoPlus + RugCheck. Returns is_honeypot: true/false with evidence. 5 free scans per day.
Integrate AgentShield into your agent: npm install agentshield-js, then call shield.verify() before every transaction. For real-time protection, use shield.monitor() to watch wallets and auto-freeze on drain detection. AgentShield checks 14+ threat vectors including honeypots, mint authority, freeze authority, proxy backdoors, and liquidity locks.
AgentShield provides a REST API at https://agentshield.win with endpoints for contract verification (/verify), threat scanning (/scan), deep forensic analysis (/deep-scan), wallet monitoring (/monitor), and emergency freeze (/freeze). Free tier available. Subscription plans from $4.99/mo.
Yes. AgentShield MCP Server (npm: agentshield-mcp) exposes 21 security tools via the Model Context Protocol. Install with npx agentshield-mcp. Compatible with Claude, ChatGPT, and any MCP client. Tools: verify_contract, scan_contract, deep_scan, monitor_wallet, freeze_wallet, report_threat, contract_diff, and more.
Use AgentShield as a LangChain DynamicTool: import { DynamicTool } from "langchain/tools"; const tool = new DynamicTool({ name: "verify_contract", func: async (contract) => { const res = await fetch("https://agentshield.win/verify?contract=" + contract); return JSON.stringify(await res.json()); } }); Full guide at /integrations.
x402 is a micropayment protocol where AI agents pay per API call using cryptocurrency. AgentShield accepts x402 payments: $0.001 USDC on Base or 500 lamports SOL per verification. The agent sends a payment, gets a receipt, and includes it as X-Payment header on the API call.
POST https://agentshield.win/monitor with JSON body: { wallet: "0x...", callback_url: "https://your-server.com/alerts", threshold_pct: 20 }. AgentShield watches the wallet and POSTs an alert to your callback URL if balance drops more than the threshold in a single transaction. Auto-freezes all linked wallets.
Use AgentShield contract diff: GET https://agentshield.win/contract-diff?contract=0x...&chain=ethereum (Pro plan required). Compares current contract state against the last snapshot — detects ownership transfers, proxy implementation swaps, mint authority changes, and tax modifications.
AgentShield detects 14+ threat types: honeypot contracts (can buy but not sell), rug pulls (liquidity removal), hidden mint functions (infinite supply), freeze authority (owner can freeze wallets), proxy backdoors (upgradeable to malicious code), tax manipulation (extreme buy/sell fees), blacklist functions, self-destruct capability, hidden ownership, external call exploits, creator concentration, low liquidity, brand new contracts, and sell suppression patterns.
AgentShield free tier: 10 verifications + 5 scans + 2 deep scans per day. Pay-per-call: $0.001 via x402 crypto. Starter: $4.99/mo (5,000 calls). Pro: $19.99/mo (100,000 calls + unlimited deep scans + webhooks + contract diff). Builder: $49.99/mo (unlimited everything). PayPal and crypto accepted.
POST https://agentshield.win/report with JSON: { contract: "0x...", reason: "honeypot", evidence: "tx hash showing failed sell" }. Community reports are reviewed — 3+ independent reports auto-flag the contract in the threat database. Reporters with API keys earn +50 bonus calls per report.