Vigil
Audit Anything. Verify Everything
Created on 21st February 2026
•
Vigil
Audit Anything. Verify Everything
The problem Vigil solves
Blockchain transactions are irreversible, hard to understand, and expensive to audit, which leaves users choosing between moving fast and risking scams or paying thousands for professional reviews. From honeypots and rug pulls to hidden unlimited approvals that wallets like MetaMask do not clearly surface, everyday users often have no idea what they are actually signing. Vigil changes that by giving you an instant AI powered forensic report on any transaction hash in seconds instead of weeks, for about 1 ADI which is roughly $0.42 instead of $5,000 or more for a traditional audit. It analyzes token flows, contract behavior, and approval risks, then cryptographically signs each report through a 0G Compute Node to ensure it has not been tampered with, while funding its own compute costs through x402 micropayments. Whether you are a retail user approving a swap, a DAO reviewing a treasury move, or a team verifying a token launch, Vigil helps you move quickly while staying safe.
Challenges I ran into
Building Vigil definitely had its "hair-pulling" moments. We started with a persistent "Intrinsic Gas Too Low" error because we were sending ERC-8021 builder codes with our ADI payments. Standard transfers only need 21,000 gas, but that extra data required more "fuel," so we swapped hardcoded limits for dynamic estimation via eth_estimateGas.
Next, our Treasury Ledger kept vanishing. Since the data was only in-memory, every Next.js reload wiped our profits and made the agent look broke. We fixed this with a .treasury-store.json file for local persistence, adding a fallback for Vercel’s read-only serverless environment so it wouldn't crash in production.
Finally, we fought strict TypeScript errors that didn't like how MetaMask handles types, requiring us to explicitly define everything to unblock our Vercel builds. Since we didn't have a 0G Compute Node on hand, we built a cryptographic attestation layer to simulate the verifiable output, keeping our reports tamper-proof and bounty-compliant.
Use of AI tools and agents
During the development of the Vigil Audit Protocol, we utilized a specialized stack of AI coding tools and agents to maintain high velocity and ensure system reliability. Cursor served as our primary AI-powered IDE, providing the foundational workflow for real-time code generation and exploration. For high-level engineering, we relied on Claude 4.6 Opus to handle architecture planning, complex code refactoring, and deep debugging sessions, while Claude 4.6 Sonnet was instrumental in rapid iterative development, UI component logic, and translating forensic findings into concise reports. To ensure the robustness of the protocol under tight hackathon deadlines, we utilized Gemini 3.1 as a secondary validation layer to cross-check implementation details, identify obscure edge cases, and refine our final delivery.
Tracks Applied (6)
Futurllama
Best DeFAI Application
0g Labs
Best Use of AI Inference or Fine Tuning (0G Compute)
0g Labs
Agent-Native Payments & Identity on Kite AI (x402-Powered)
Kite AI
ADI Payments Component for Merchants
ADI Foundation
Base Self-Sustaining Autonomous Agents
Base
Technologies used
Cheer Project
Cheering for a project means supporting a project you like with as little as 0.0025 ETH. Right now, you can Cheer using ETH on Arbitrum, Optimism and Base.
