Inspiration
Traditional security tools only tell you what's broken after deployment. We wanted to flip this - what if your AI coding assistant knew your organization's actual security vulnerabilities and could prevent them during development? Instead of playing security whack-a-mole, developers could write secure code from day one.
What it does
SecureCodeAI is an AI pair programmer that integrates with Wiz's security platform to provide context-aware code suggestions. When you're writing infrastructure code, it:
- Analyzes your organization's real security vulnerabilities via Wiz MCP
- Generates code that avoids patterns that have caused issues in your environment
- Provides real-time warnings when your code matches existing vulnerability patterns
- Learns from your deployment outcomes to improve future suggestions
How I built it
- Frontend: VS Code extension with React-based chat interface
- Backend: Node.js server orchestrating between claude and Wiz MCP tools
- AI Pipeline: Custom prompt engineering that injects real security context from Wiz
- Integration: Wiz MCP server providing live security data (vulnerabilities, threats, configurations)
- Learning Loop: Feedback mechanism that correlates code suggestions with post-deployment security scan results
Challenges I ran into
- understanding the aws bedrock agentcore interface and understanding the wiz mcp server
🏆 Accomplishments that I proud of
- Successfully integrated live security data into code generation for the first time
- Deployed an mcp server on AWS build agentcore for the first time
📚 What I learned
- Security context dramatically improves AI code quality when properly integrated
- Developers are more receptive to security guidance when it's personalized to their environment
- Real-time security data can be effectively used for proactive rather than reactive security
- The key is making security invisible to developers while still being comprehensive
🚀 What's next for SecureCodeAI
- Multi-IDE Support: Expand beyond VS Code to JetBrains, Vim, etc.
- Team Learning: Share security insights across development teams
- Compliance Integration: Automatically ensure code meets SOC2, HIPAA, etc. requirements
- Security Metrics: Track security improvement trends across development cycles
- Enterprise Deployment: Scale to handle large organizations with complex security policies
*Built with: Wiz MCP, AWS bedrock agentcore, claude
Built With
- agentcore
- amazon-web-services
- bedrock
- mcp
- nextjs
- wiz

Log in or sign up for Devpost to join the conversation.