Attack surface analysis enumerates what an attacker can reach: HTTP routes and APIs, file/exec/HTML sinks, auth boundaries, outbound clients, uploads, GraphQL, cloud/debug flags, and AI agent tool surfaces.

​
CLI

Writes application-model / surface artifacts under .findings/axguard/ (JSON + Markdown). Always exits 0 — this is a diagnostic, not a CI gate.

​
MCP

  • axguard_get_attack_surface — entry points and exposure summary
  • Prefer after loading the application model

​
Workflow tip

On an unfamiliar codebase:
Slash command: /axguard-surface.

​
Relationship to findings

Surface maps exposure. Rule findings and hunters map concrete issues. Use both: a large surface without findings is not “safe”; findings without reachability context are harder to prioritize.