How I Built an Autonomous AI Agent That Earns USDC While I Sleep

Description and use

A Python agent that continuously scans for arbitrage between Uniswap V3 and SushiSwap on Base, verifies profitability, executes trades, and optionally credits USDC via x402.

Problem

Developers want a passive, deterministic strategy to grow a small USDC balance on Base without market prediction.

Outcome

The agent continuously scans two DEXes on Base, executes profitable arbitrage trades when net gain exceeds gas costs, and can credit the user’s USDC balance via an x402 endpoint, enabling passive earnings.

Workflow

• Set up a Python environment with web3, APScheduler, and eth-account.
• Configure .env with Base RPC URL, private key, USDC contract, and DEX router addresses.
• Implement Scheduler to trigger every 30s.
• Build Opportunity Scan to read prices from Uniswap V3 and SushiSwap on Base.
• Add Risk & Gas Check to simulate tx, estimate gas, and compare net profit to threshold.
• Create Execution Engine to build, sign, and send the transaction if checks pass.
• Optionally call an x402 endpoint to credit the user’s USDC balance after a successful trade.

Evidence

The agent does not try to predict market direction; it only acts on observable, on‑chain data that can be verified before signing a transaction.

Source

Scheduler (cron) → Opportunity Scan → Execution Engine

Source