Kyvvu

OWASP agentic · ASI07

ASI07: Insecure Inter-Agent Communication

Multi-agent systems exchange messages over APIs, message buses, shared memory and protocols such as MCP and A2A. When those exchanges lack authentication, integrity checks or semantic validation, attackers can intercept, spoof, replay or alter messages. This can inject instructions, confuse authority or misdirect work between agents.

What it looks like

  • A man-in-the-middle on an unencrypted channel injects hidden instructions into messages between agents.
  • An attacker registers a fake peer in an agent discovery service and receives privileged coordination traffic.
  • Old delegation or emergency messages are replayed, or agents are forced down to a weaker protocol mode.

What OWASP recommends

  • Encrypt agent channels end to end with per-agent credentials and mutual authentication.
  • Sign messages, protect against replay with nonces and timestamps, and validate them against typed, versioned schemas.
  • Pin allowed protocol versions and reject downgrades.
  • Authenticate discovery and use attested registries with signed agent cards.

How Kyvvu's policies address it

No policy in the open library is mapped to ASI07 yet. See the policy library for what it covers today.

Based on the OWASP Top 10 for Agentic Applications for 2026 by the OWASP GenAI Security Project, licensed CC BY-SA 4.0. The names are OWASP's; the summaries are ours.

Your agents

Is ASI07 a risk for yours?

For security teams

Request a demo

We walk through your agents' riskiest paths, and the policies that close them.

For developers

Build the path yourself

Recreate the attack in the playground and see which policies stop it.