The application model is AXguard’s structured understanding of a codebase: frameworks, routes/entrypoints, components, sinks, and related metadata. Later stages (surface, dataflow, attack paths, twin) consume this model instead of re-guessing the app shape.

CLI

axguard audit soft-runs the surface / model phase and writes the same artifacts alongside vulnerability reports.

MCP

  • axguard_get_application_model — load the model for the current project (soft if unavailable)
  • axguard_get_project — workspace root / MCP config summary (not a security verdict)

What it is for

  • Ground threat modeling and attack-surface maps in observed structure
  • Give dataflow and hunters a shared graph of entrypoints and sinks
  • Keep reports honest when modeling is incomplete (limited analysis, not a fake clean bill)

Limits

The model is best-effort static understanding. Missing adapters or exotic frameworks yield a partial model — diagnostics continue rather than inventing coverage.