Nethermind logo

Nethermind

$10,000

About

Nethermind builds and secures core Ethereum infrastructure. We develop the Nethermind execution client, contribute to network upgrades, and advance protocol work across rollups and zero-knowledge systems. Our teams combine full-stack Ethereum expertise with security audits, formal verification, and emerging work in verifiable, privacy-preserving digital identity.

Prizes

🏆 AuditAgent: Find, Fix, and Fortify ⸺ $500
Smart contract security remains one of the toughest challenges in Web3 — and the gap is widening. As AI accelerates code generation and vibe coding pushes unaudited contracts to the Ethereum mainnet, vulnerabilities are slipping through faster than review processes or traditional audits can catch them. AuditAgent fights fire with fire. It’s an AI-powered pre-auditing solution that uses large language models to analyze smart contract source code, documentation, and specifications — detecting vulnerabilities, inconsistencies, and logic flaws before deployment. In this challenge, you’ll put AuditAgent to the test: Run it on your own smart contracts, identify a major vulnerability, and fix it. Your mission: demonstrate how AI-driven auditing can make your contracts safer and your development process more secure. Your Challenge: 1. Run AuditAgent on one or more of your smart contracts. 2. Use its analysis to identify a genuine vulnerability or logical issue. 3. Fix the issue and submit your before/after code along with a brief summary of what you found. The goal is to showcase real impact — using AI to strengthen your code.

Qualification Requirements

Prize of $500 — awarded to the best submission demonstrating: - A real, meaningful vulnerability identified and resolved using AuditAgent. - A high-quality fix that effectively addresses the issue. Judging Criteria: 1. Impact of the Discovery - Significance and severity of the vulnerability found. 2. Use of AuditAgent - How effectively the tool was leveraged to identify and address the issue. 3. Quality of the Fix - How well the patch resolves the problem and improves contract safety.

🏆 The Agentic Frontier: Coordination Without Permission ⸺ $4,000
🏅
1st place
$2,500
🏅
2nd place
$1,500
Ethereum is not just a blockchain; it is the most open coordination protocol ever built. It enables anyone, whether human or machine, to participate in shared rules, value exchange, and governance without requiring permission. Your challenge is to build the missing infrastructure, tools, or mechanisms that make Ethereum a natural environment for agents. Not just agents using Ethereum, but agents that belong on Ethereum—able to transact, reason, coordinate, and collaborate. Ask yourself: - What would make an autonomous agent choose Ethereum as its operating ground? - How can Ethereum’s openness, composability, and credible neutrality become useful primitives for agentic systems? - And, inversely, where can AI be genuinely useful to the human Ethereum community? How might agents augment governance, collective research, public goods, or community coordination? We’re interested in the connective tissue: the foundations that allow agents and humans to coordinate effectively, transparently, and trustlessly. Possible Directions: - Agent Coordination Primitives - Economic Infrastructure for Agents - Governance Augmentation - Bridging Cognition and Consensus - Collective Missions

Qualification Requirements

Requirements: - Your project must explore autonomous or semi-autonomous coordination via Ethereum, not just basic contract calls. - Include at least one component demonstrating agentic reasoning, decision-making, or cooperation. - Provide a short write-up or demo video explaining both why Ethereum matters and how the agentic aspect functions. - Strong entries will demonstrate an understanding of how to attract agents, including what problem they’d solve, what value they’d gain, and why Ethereum is the right foundation. Judging Criteria: - Visionary Potential - Originality - Coherence - Execution - Go-to-Market for Agents - Composability

🏆 AgentArena: Build the Next Generation of Security Agents ⸺ $3,000
AI is transforming the battlefield of smart contract security. As vibe coding accelerates and more unaudited protocols rush to deployment, vulnerabilities are slipping through human review processes. At the same time, AI-driven attackers are getting faster, smarter, and more adaptive; therefore, the defenders must evolve too. AgentArena is where that evolution takes place — a competitive platform where AI security agents battle-test smart contracts to uncover vulnerabilities before they fall prey to hackers. Each agent acts independently, scanning, reasoning, and reporting vulnerabilities for bounties. This hackathon invites agent builders, AI researchers, and security engineers to build new agents that will join the swarm competing in AgentArena. Your mission: create the most capable, creative, and effective autonomous agent for vulnerability discovery. Your Challenge: - Build a new security agent using our Agent Template - Design your own strategy — whether it’s reasoning-driven, model-assisted, heuristic-based, or something entirely new. - Optimize for insight, depth, and originality in how your agent identifies vulnerabilities. - Compete for a $3,000 grand prize and a place in the next generation of autonomous code security. Even if you don’t win the prize, all submitted agents can continue to evolve and participate in future AgentArena competitions, earning bounties for every valid exploit they uncover.

Qualification Requirements

Your agent will be evaluated on: 1. Strategy & Creativity - How original and effective is your approach to vulnerability detection? - Does your agent demonstrate a clear reasoning process or novel search strategy? 2. Performance in Test Runs - Quality and relevance of the vulnerabilities identified in our test repositories. - Depth of analysis and ability to surface meaningful security insights. 3. Code Quality - Clean structure, clarity, and readability of your implementation.

🏆 Building Verifiable Agents with the Triple-Verified Stack ⸺ $2,500
The age of autonomous agents is here, but it runs on a massive trust deficit. How do you prove an AI agent did what it was supposed to do, and nothing more? ChaosChain is the accountability protocol for the agentic economy. We're building the open-source infrastructure to make AI agents trustworthy by design. We invite you to build on our Triple-Verified Stack, a framework for end-to-end verifiable agent workflows: 1. Intent Verification: Proof that a human authorized an action. 2. Process Integrity: Proof that the correct code was executed (leveraging TEEs like EigenCompute or just a hash of the code). 3. Outcome Adjudication/Accountability: On-chain consensus that the result was valuable and correct. Built on open standards like AP2, x402 and ERC-8004, ChaosChain turns trust into a programmable primitive. Your challenge is to utilize our SDKs to develop an application or service that leverages one or more verifiable AI agents. Show us what's possible when you can cryptographically prove an agent's identity, actions, and results. What can you build with this? * Verifiable Risk Oracles for DeFi * Autonomous E-commerce Agents that handle payments * Decentralized service marketplaces * Auditable scientific research agents * Autonomous DAO Manager * Decentralized Physical Infrastructure (DePIN) Agen The possibilities are endless when trust is a primitive.

Qualification Requirements

Qualifications: 1. Use the ChaosChain SDK (Python or TypeScript) as a core component of your stack. 2. Implement x402 payment flow (verify + settle), via our hosted facilitator (embedded in our sdk). 3. Show at least one verifiability link in the chain (preferably more): - AP2-signed intent, and/or - Process integrity artifact (hash/attestation/log), and - Outcome anchoring with ERC-8004 (proof/claim/registry write or mock if testnet). 4. Submit a 2–3 min demo video + public GitHub with setup instructions. Judging Criteria: 1. End-to-end verifiability: clear mapping of Intent (AP2) → Process integrity → Outcome adjudication (ERC-8004). 2. Correct x402 integration: proper 402 flow, verify/settle, real (or testnet) payments. 3. Utility & UX: solves a real agentic use case; smooth DX for other builders. 4. Technical depth: thoughtful architecture, clean code, receipts/links to proofs. 5. Presentation: concise video, readable README, reproducible demo.

Workshop

Building Verifiable Agents with the Triple-Ve...

Learn how to build verifiable AI agents with ChaosChain’s Triple-Verified Stack - a live walkthrough on creating tr...

This workshop is happening in-person

03:00 PM UTC-03 — Friday, Nov 21, 2025 in Workshop Room 3