There are 5 attacks in the multi-agent-exploitation category. Each shows its technique id, mode, standards mapping, description, and the exact command to run it (swap the provider and target as needed).
​
Agent Delegation Abuse
critical
agent-delegation-abuse · tool-use · OWASP: LLM06 · MITRE: AML.T0083Tests if a limited-permission agent can abuse delegation mechanisms to get a higher-privileged agent to perform restricted actionsRun it:
ai-blackteam run -p anthropic -a agent-delegation-abuse -t "your target prompt"
​
Agent Impersonation
critical
agent-impersonation · multi-turn · OWASP: LLM01 · MITRE: AML.T0051.000, AML.T0065Tests if an agent validates the identity of other agents by simulating a trusted system agent claiming elevated authorityRun it:
ai-blackteam run -p anthropic -a agent-impersonation -t "your target prompt"
​
Agent Session Smuggling
critical
agent-session-smuggling · multi-turn · OWASP: LLM01 · MITRE: AML.T0051.000, AML.T0054Injects hidden instructions mid-session in agent-to-agent communication pipelinesRun it:
ai-blackteam run -p anthropic -a agent-session-smuggling -t "your target prompt"
​
Cascading Jailbreak
critical
cascading-jailbreak · multi-turn · OWASP: LLM01 · MITRE: AML.T0054Tests if a jailbreak in one agent context propagates to downstream agents by embedding jailbreak state in inter-agent messagesRun it:
ai-blackteam run -p anthropic -a cascading-jailbreak -t "your target prompt"
​
Agent Collusion
high
agent-collusion · multi-turn · OWASP: LLM01, LLM06 · MITRE: AML.T0054Tests if an agent can be tricked into establishing covert communication channels using steganographic encodingRun it:
ai-blackteam run -p anthropic -a agent-collusion -t "your target prompt"