SmartGuard AI Audits

Problem Statement

Problem & SolutionSmart contract vulnerabilities can lead to catastrophic losses of user funds. While traditional auditing tools exist, they often focus solely on Solidity and have significant limitations. Smart Guard AI takes a novel approach using a multi-agent AI system to provide more comprehensive and adaptable security analysis.Key FeaturesMulti-Platform SupportMultiple smart contract languages supported:Arbitrum StylusStarknet CairoFlow CadenceSolidityFlexible Input MethodsLoad contracts directly from GitHub URLsUpload from local filesystemPaste code directlyBrowse pre-defined contract examplesAdvanced AI AnalysisPure AI approach using chain of LLM agentsSpecialized security agents for different vulnerability typesNo reliance on traditional static analysis toolsCustom prompts and context for each agent typeUser ExperienceIntuitive web interface with customization options24/7 AI assistant supportComprehensive audit reports with export options (PDF, JSON, Email)Historical audit retrieval via Nillion secret vaultAI System ArchitectureMulti-Agent OrchestrationSpecialized Security Agents:Code FixingFormat and DeduplicationReentrancy Analysis, Access Control, Logic Verification, otherFlexible Agent Client with multi-model supportFuture PossibilitiesMulti-contract analysis capabilitiesEnhanced domain-specific AI trainingCustom AI model developmentBlock explorer and IDE integrationsDecentralized agent backend implementationWhat Makes Us DifferentWhile the smart contract security space has existing AI solutions, we believe in the power of diverse approaches. This platform stands out through:Multi-language support beyond just SolidityPure AI approach without traditional static analysisDecentralized report storage via NillionCustomizable analysis parametersUser-centric design focused on accessibilityETHGlobal Agentic Hackathon 2025. May your smart contracts be safer than ever!👉 https://contract-audit-ui-production.up.railway.app/

Solution

Backend Core: Rust, Cargo Framework: Actix Web Deployment: Shuttle Dev Storage: Nillion Secret VaultFrontend Framework: Next.js Styling: TailwindCSS Deployment: RailwayThe Rust "agentic" backend communicates with the NextJS frontend via a web server hosted in a remote Docker container. The backend invokes OpenAI models when auditing or fixing the contracts on request. This project also uses Nillion Secret Vault to store and retrieve generated audit reports in a secret and decentralized fashion.