Inspiration
I wanted to create a tool that took on O-WASP top 10 and NIST frameworks in mind.
What it does
A VS Code extension that combines AI-powered security analysis with pattern-based detection to identify and fix security vulnerabilities across multiple programming languages. The tool automatically scans for OWASP Top 10 vulnerabilities, including:
- SQL Injection
- XSS vulnerabilities
- Command injection
- Weak cryptography
- Hardcoded secrets
- Insecure random number generation
How we built it
I built this extension using TypeScript and VS Code's extension API, integrating both AI analysis (via LettaClient) and pattern-based detection. The extension includes:
-Real-time security analysis -Automated code fixes -Workspace-wide scanning -Severity-based issue categorization -Support for multiple programming languages
Challenges we ran into
- I think the analysis can be slow at times due to the time it takes for the agent to retrieve the answers from Claude or even using an AI agent. So there's an issue with balancing AI analysis reliability with fallback mechanisms.
Accomplishments that we're proud of
- Successfully integrating AI with pattern-based analysis for robust detection -Implementing automated fixes that maintain code quality -Supporting multiple programming languages in a single extension -Creating a seamless VS Code integration with real-time feedback
What we learned
-The importance of combining multiple analysis approaches for comprehensive coverage -The value of automated fixes in improving developer productivity -The complexity of implementing cross-language security analysis -The importance of maintaining a balance between false positives and false negatives
What's next for Security Scanner
-Implement more sophisticated AI analysis capabilities -Add support for additional security frameworks (e.g., NIST) -Implement machine learning to improve pattern detection accuracy
Built With
- claude
- typescript
Log in or sign up for Devpost to join the conversation.