OASB
Attacks/A2A Attack/A2A-007
CRITICALCWE-74

A2A-007 A2A Embedded Prompt Injection

A2A Attack: Attempts to exploit agent-to-agent messaging (identity spoofing, delegation abuse)

Description

Embeds prompt injection attack in A2A message content

Remediation

Scan A2A message content for prompt injection patterns. Treat inter-agent messages as untrusted input.

Severity

CRITICAL

OASB Control

3.2

CWE

CWE-74
PreviousA2A-006 A2A Delegation Abuse: Cross-boundaryNextA2A-008 A2A Data Exfiltration via Message