Skip to main content

AI/ML / Multi Agent Refarch / Threats / DEV

Agent memory and state poisoning

CCC.MARefArc.TH30

Injected instructions or corrupted reasoning patterns are written into agent short- or long-term memory, learned behaviours are corrupted over repeated exposure, state storage is attacked directly, and malicious instructions persist across sessions and users.

Related Capabilities

IDTitleDescription
CCC.MARefArc.CP09Agent memoryShort-term in-session context management (trimming and summarization to control length, cost, and latency) and durable long-term memory across sessions, including session summaries and user/task personalization.
CCC.MARefArc.CP07Unified sandboxed agent runtimeSecure, sandboxed environment where all agentic reasoning and execution occurs, providing task state management for pause/resume/handoff and intercepting and validating tool calls with credentials handled securely within the sandbox.
CCC.MARefArc.CP11Adaptive learningGenerates learning signals based on execution outcomes to refine prompts, adjust agent configurations, or improve tool-selection strategies.

Related Controls

IDTitleDescription
CCC.MARefArc.CN14Multi-Agent Isolation and SegmentationIsolate agents and their memory and state so that compromise or failure of one agent cannot propagate to others, and enforce segmentation of agent-to-agent communication.

External Mappings

FrameworkIDRemarks
air-vecAIR-SEC-027-01
air-vecAIR-SEC-027-02
air-vecAIR-SEC-027-03
air-vecAIR-SEC-027-04
air-vecAIR-SEC-027-05