AG07:2026 — Insecure Inter-Agent Communication

Slide 23 · Mitigation Matrix

Which mitigations address which attack types and scenarios.

Attack / Scenario	MIT01 Auth	MIT02 Signing	MIT03 TLS	MIT04 Validation	MIT05 Replay	MIT06 Routing	MIT07 Registry	MIT08 Logging	MIT09 Anomaly
T1 — Message forgery	●	◉	◉	◉		◉	●	◉	◉
T2 — Tampering in transit		●	◉	◉				◉
T3 — Orchestrator impersonation	●	◉	●			●	●	◉	◉
T4 — Replay attack		◉			●			◉	◉
T5 — Eavesdropping			●
T6 — Queue injection	●			◉		●	◉	◉	◉

● Primary mitigation — directly prevents this attack ◉ Contributing mitigation — reduces severity or aids detection

Key insight from the matrix

MIT01 + MIT03 + MIT06 + MIT07 form the identity and routing foundation — they appear as primary controls for four of the six attack types. Every other control builds on top of this foundation.

MIT02 is the only primary control for Type 2 (tampering). No other mitigation directly prevents message tampering once the channel is authenticated. This means MIT02 (message signing) is non-negotiable if tamper-resistance is required — TLS alone is insufficient because TLS only protects transport, not the message body at application layer.

Type 5 (eavesdropping) has only one primary control: MIT03 (TLS). If TLS is absent, eavesdropping cannot be prevented by any other mitigation on this list.

← Back Quiz →