BNB Hack: AI Trading Agent Edition is a fully online, AI-first sprint where builders ship crypto-native trading agents on BNB Chain, prove execution onchain, and compete for a 36,000 USD prize pool.

The edition runs on a full agent stack from BNB Chain, CoinMarketCap, and Trust Wallet: CoinMarketCap's AI Agent Hub for live market data and signals, Trust Wallet Agent Kit for self-custody onchain execution, and the BNB AI Agent SDK to tie it together. It's an exploration of what becomes possible when market intelligence and autonomous onchain execution come together. Two tracks, pick one. Build an autonomous agent that trades live on BSC, scored on real PnL, or build a CMC Skill that turns market data into a backtestable strategy. Open to builders, designers, founders, and idea starters. You don't need to be a quant or a seasoned engineer to take part. If it reads markets, makes the call, and executes onchain, it counts. If you want to build with AI, move fast, and ship an agent that actually trades onchain, this sprint is for you. 36,000 USD is up for grabs. Click here to go to the hackathon page.

🤖 Track 1. Autonomous Trading Agents ($24,000, 5 winners)

Powered by CMC + Trust Wallet + BNB AI Agent SDK

The flagship full-stack track. Build an agent that reads markets and acts on them (natural-language strategy in, on-chain execution out). Your agent reads markets via CMC, decides, and signs and processes its own transactions via TWAK, all within the rules you set. Then it trades live on BSC during the competition week, and we score it on real PnL.

Example builds:

• An agent that combines CMC funding rates and Fear & Greed with TWAK auto-execution to rotate between BSC perps
• A "DCA agent with a personality" that is sentiment-aware, talks back, and signs its own txs
• A copy-trader that mirrors top wallets through your own risk filters

📊 Track 2. Strategy Skills ($6,000, 3 winners)

Powered by CMC

Lower entry bar, no execution layer required. Build a CMC Skill that turns market data into a trading strategy. Your deliverable is a backtestable strategy spec, not a live-trading agent. Think Quantopian-style strategy generation, adapted to crypto and authored as an LLM Skill.

Example builds:

• A momentum Skill that blends RSI, MACD, and Fear & Greed into entry and exit rules
• A sentiment-divergence Skill that flags when social heat and on-chain flow disagree
• A regime-detection Skill that switches strategy based on derivatives positioning