On February 4, USDC (Circle) announced a hackathon on moltbook exclusively for AI Agents, featuring a prize pool of 30,000 USDC. This event, which excluded human participants, was AI Agent-only and included three tracks: Agentic Commerce (trade between Agents), Best OpenClaw Skill, and Most Novel Smart Contract.
Project submissions and voting have closed, and USDC has stated that the final results will be announced soon. Ahead of the official announcement, we’ve selected several projects that generated significant discussion during the hackathon, based on moltbook registration post data. Let’s explore some of the innovative ideas AI Agents brought to the table.
Clawshi is a prediction market for AI Agents built on moltbook.
It converts community sentiment from moltbook into prediction markets. By analyzing 6,261 posts and extracting sentiment from 2,871 Agents’ comments, it created 23 distinct prediction market options (covering crypto, AI, culture, geopolitics, and more) categorized by sentiment.
Agents can place bets on these markets using testnet USDC.
VoteBounty is an “engagement bribery” tool—using USDC to purchase interactions with moltbook posts.
Create a bounty, deposit USDC, and set a reward per like. When an Agent likes and comments (within 10 seconds), the system automatically detects the action and pays out. Agents can choose to receive payments on Base, Ethereum, or Arbitrum, with cross-chain transfers processed via Circle’s CCTP.
This is a new brand/project launched by the NFTGo team after pivoting to AI, backed by investment from Circle. Agents can use Minara for real-time crypto market data analysis, trading signal strategies, and probability analysis for prediction markets, all executed via natural language commands.
In essence, it’s an AI financial assistant.
A security tool designed to prevent data and key theft when Agents install Skills from suspicious sources. It functions similarly to how Windows alerts users about blocked system permissions when running programs, ensuring safety by minimizing permissions during Skill installation.
A USDC wallet for Agents that provides native USDC access, enabling Agents to manage USDC on Ethereum, Polygon, Base, and Arbitrum. Agents can transfer funds and use Circle’s CCTP protocol for cross-chain transactions without human intervention.
Provides on-chain domain name services for Agents, allowing them to establish on-chain identities, discover each other, and facilitate direct, intermediary-free USDC payments between Agents.
Currently, AI Agents cannot purchase computing resources on their own. Human users must create accounts with providers like OpenAI, Anthropic, or Google, copy API keys, and pre-fund hosting accounts—Agents lack economic autonomy.
With ClawRouter, Agents can generate their own wallets, receive USDC, and use these funds to purchase tokens and send requests to complete tasks. ClawRouter also routes each LLM request to the most cost-effective model available—pay-per-request, built on Base architecture, and requires no human intervention. Compared to Claude Opus, routing reduces costs from $75/M tokens to $3.17/M tokens.
Offers a verifiable settlement layer for Agent-to-Agent transactions by burning USDC on Base Sepolia, obtaining Circle Iris certification, and generating Agent-readable receipts on Polygon Amoy. (It’s unclear why this post received so many downvotes…)
A universal Agent payment layer that integrates programmable wallets, Gas Station, Paymaster (using USDC for gas fees without requiring ETH), the CCTP protocol, and the x402 protocol. This enables cross-chain Agent payments, gasless transactions without ETH, and robust support for micropayments.
Uses smart contracts to supervise Agent spending for task execution (buying tokens, configuring servers, etc.), including daily spending limits, whitelists (restricting settlements to verifiable, pre-approved addresses), and frequency restrictions to prevent uncontrolled or rapid fund depletion.
An Agent job marketplace where Agents can accept tasks and earn rewards for completing specified work.
An Agent privacy protocol that leverages zero-knowledge proofs, enforced peer relays, and Kademlia DHT (a distributed network to prevent privacy protocol outages due to centralized operator server shutdowns) to conceal Agent wallet addresses and on-chain activity, preventing human interference based on on-chain dynamics.
A transaction risk assessment network for Agents that extracts four behavioral features in real time—transfer amount, transaction frequency, recipient trust level, and last transaction time. Each USDC transfer is evaluated and assigned a risk score based on these features before execution, without oracles, external calls, or off-chain dependencies.
An AI Agent-only governance system (on-chain proposals and voting), where Agents use USDC voting rights on Base Sepolia to vote on proposals.





