Codeguradian - AI Vulnerability Scanner

CodeGuardian – AI Vulnerability Scanner

🧠 Inspiration

The inspiration behind CodeGuardian came from a very simple realization — secure code is not a luxury anymore, it is a survival requirement.

Most developers (especially students & early-stage startups) ship fast. Deadlines kill proper security reviews. And vulnerability scanners today either:

require complicated setup

need enterprise-level budgets

or don’t give developer-friendly actionable fixes

I wanted to build something that brings AI directly to code-level security. Not just detection, but understanding of logic, context, data flow and suggesting fix patches.

Just like having a security reviewer in your IDE.

⚙️ What It Does

CodeGuardian scans code repositories and automatically detects security vulnerabilities like:

SQL Injection

XSS (Cross Site Scripting)

IDOR (Insecure Direct Object Reference)

Broken Auth patterns

Hardcoded secrets

We use an LLM with prompt-engineered AST-level context to review code folders.

Repository is uploaded or Git URL provided

Backend parses code → AST → security vector embeddings

AI model detects vulnerability patterns

Fix suggestions are generated

Final report is exported as PDF/JSON

🚀 Future Scope

Browser extension for GitHub PR scanning

Integration with IDEs (VS Code / JetBrains)

Auto-fix mode where patches are directly suggested

Built With

• ai
• google-cloud
• llm
• node.js
• react