Agent SecuritySimulationSupabaseEducation
AI Security Navigator
Interactive periodic-table style navigator for AI security learning, design recommendations, and safe simulations with a constrained execution model.
Problem / Scope
AI Security Navigator is a learning and design surface for AI security concepts. The portfolio version emphasizes a constrained simulation model: users can explore templates and scenarios, but they should not be able to execute arbitrary code.
Architecture
- Frontend on Vercel
- Supabase for content, sessions, and persisted traces
- Optional VPS worker for heavier simulation work
- Safe simulation templates rather than arbitrary user-authored execution
Key Workflows / What It Proves
- Learn mode: browse concepts in a structured visual system
- Design mode: get recommendations for controls and patterns
- Simulate mode: walk through predefined scenarios and inspect traces
Limitations
- Simulation fidelity is bounded by the predefined templates
- Browser-only fallback mode is narrower than the optional hybrid deployment
- The published demo is intended to prove interaction and safety constraints, not production-scale coverage
Evidence Pack
E-NAV-001
Navigator table view
Screenshot of the main table and detail panel.
E-NAV-002
Design mode recommendations
Published screenshot showing recommendation output and framing.
E-NAV-003
Simulation session trace
Trace and events screenshot demonstrating safe, template-based simulation flow.
E-NAV-004
Markdown report export
Snapshot of the export artifact copied into the hub evidence pack.
E-NAV-005
Repo commit permalink
Pinned source reference for the demo build.