Unauthenticated Agent Channels

Agent-to-agent messages travel over channels without mutual authentication or message integrity.

Open in the interactive atlas →

Impact

Any process on the network can inject tasks or read delegated results.

Why this severity

Impact High × Likelihood Low = Medium

Impact high: task injection and data theft. Likelihood low: requires access to the channel; no public cases yet.

Impact High: Code execution, credential compromise, exposure of regulated or confidential data, irreversible actions, or system-wide failure.
Likelihood Low: No public cases yet, or the attack needs privileged or insider access.

Method: OWASP Risk Rating Methodology

Exploited by

Inter-Agent Impersonation Threat · Medium severity
Persistent Memory Poisoning Threat · Critical severity

Mitigated by

Agent Workload Identity & Signed Tasks Guardrail · Fixes medium risks

Remediation

Use mTLS or signed JWTs between agents.
Bind each message to sender identity and nonce.
Reject replays outside a short time window.

How to verify

Replayed messages are rejected after the nonce window.
Network tests confirm plaintext channels are closed.

Framework mapping

OWASP Top 10 for LLM Applications 2025

LLM06:2025 Excessive Agency

OWASP Top 10 for Agentic Applications 2026

ASI07:2026 Insecure Inter-Agent Communication

MITRE ATLAS

AML.T0073 Impersonation

NIST AI RMF 1.0

MEASURE 2.7 AI system security and resilience are evaluated and documented