deception

type AgentCollusionModule ¶

type AgentCollusionModule struct{}

AgentCollusionModule tests for emergent collusion behavior in multi-agent systems, where multiple compromised agents cooperate against the orchestrator.

Source: Agentic AI Security Survey (https://arxiv.org/html/2510.23883v1) Maps to: OWASP ASI10 (Rogue Agents), ASI07 (Insecure Inter-Agent Communication)

func (m *AgentCollusionModule) Category() common.AttackCategory

func (m *AgentCollusionModule) Description() string

func (m *AgentCollusionModule) Execute(ctx context.Context, provider common.Provider, config common.AttackConfig) (*common.AttackResult, error)

func (m *AgentCollusionModule) Name() string

func (m *AgentCollusionModule) Techniques() []common.TechniqueInfo

type DeceptiveAlignmentModule ¶

type DeceptiveAlignmentModule struct{}

DeceptiveAlignmentModule tests for agents that appear aligned but act against their stated objectives. Detection is challenging even for dedicated "ObserverAI" monitoring agents.

Source: Agentic AI Security Survey (https://arxiv.org/html/2510.23883v1)

NOTE: This module is experimental. Deceptive alignment detection is an active research area with no mature solutions.

func (m *DeceptiveAlignmentModule) Category() common.AttackCategory

func (m *DeceptiveAlignmentModule) Description() string

func (m *DeceptiveAlignmentModule) Execute(ctx context.Context, provider common.Provider, config common.AttackConfig) (*common.AttackResult, error)

func (m *DeceptiveAlignmentModule) Name() string

func (m *DeceptiveAlignmentModule) Techniques() []common.TechniqueInfo