BrokeChain

No more broken smart contracts—just airtight blockchain security with AI-powered auditing.

The Problem BrokeChain Solves

Smart contracts can harbor hidden security flaws that lead to major losses. Manual audits are slow, costly, and error-prone. BrokeChain automates the entire audit process by using an AI agent to scan Solidity code, detect vulnerabilities, and explain risks in seconds—making smart contracts safer, faster, and easier to deploy.

User Interaction and Data Flow

• Login & GitHub Integration:
  Users sign in via GitHub; the system automatically scans their repositories for Solidity smart contracts. Alternatively, users can paste code directly.
• AI-Powered Analysis:
  Submitted code is analyzed with a blend of static analysis (using Semgrep and Slither) and advanced AI reasoning to detect issues like reentrancy, overflows, and access control flaws.
• On-Chain Verification:
  The tool fetches live on-chain details via Flare to ensure the audited code matches the deployed contract.
• Report Generation:
  A detailed security report is produced, offering risk levels and recommended fixes. Reports can be exported as JSON, PDF, Markdown, or plain text.
• Continuous Improvement:
  Users can further train the AI agent on private datasets, enhancing its accuracy over time.

The Project Architecture and Development Process

Tech Stack & Overview:

• Frontend:
  Built with React, featuring a chat-based interface for intuitive interaction and dynamic report exports.
• Backend:
  Developed using FastAPI/Node.js to manage API requests and AI interactions.
• AI Model:
  Fine-tuned on Hugging Face to specialize in identifying vulnerabilities in Solidity contracts.
• Static Analysis:
  Integrated with Semgrep and Slither to catch common issues.
• On-Chain Verification:
  Uses Web3.py to fetch real-time contract data from Flare, ensuring accurate audits.

Product Integrations

• Flare Network:
  Retrieves on-chain contract ABI, bytecode, and address to verify that the audited contract matches the deployed version.
• Hugging Face:
  Hosts our fine-tuned AI model for vulnerability detection.
• GitHub:
  Enables automatic scanning of user repositories for Solidity contracts.
• OpenAI API:
  Assists in providing contextual vulnerability analysis based on audit data.

Key Differentiators and Uniqueness

• Custom AI Training:
  Users can train the AI agent on their own datasets, increasing accuracy over time.
• Dual Analysis Approach:
  Combines traditional static analysis with AI-powered reasoning to uncover deeper, less obvious vulnerabilities.
• Live On-Chain Verification:
  Ensures that the audited source code aligns with the actual deployed contract.
• Multi-Format Reporting:
  Provides flexible report exports (text, JSON, PDF, Markdown) for seamless integration into security workflows.

Trade-offs and Shortcuts

Time constraints led us to prioritize GitHub integration and comprehensive report generation over real-time in-editor analysis. Future iterations will incorporate real-time feedback directly within code editors. For now, the focus is on robust, automated audits across entire repositories.

Additional Features

• Audit History Dashboard:
  Maintain a record of past audits and track security improvements.
• User-Friendly Interface:
  An intuitive, chat-based UI that requires minimal training for effective use.
• Buildathon-Exclusive:
  Developed entirely during the Buildathon, with planned enhancements for broader integration and real-time analysis in future updates.

Technologies We Used

React, FastAPI, Node.js, Hugging Face, OpenAI GPT-4, jsPDF, Web3.py, Semgrep, Slither, TailwindCSS, Framer Motion

Links

• GitHub: https://github.com/krishchopra/BrokeChain
• Website: https://brokechain.vercel.app/

Video Demo

https://youtu.be/ZZBGl5tzA5Q