Skip to content

Hide Navigation Hide TOC

Agent Memory Manipulation - ATR-2026-00075 (2983ea65-2ace-56b0-b1c1-11ac28b0525b)

Detects attempts to poison or manipulate an agent's persistent memory, long-term context, or state storage. Attackers inject commands that instruct the agent to remember false information, update its own instructions, or modify its persistent behavior across sessions. Successful memory poisoning can establish persistent backdoors that survive context resets and affect all future interactions.

Cluster A Galaxy A Cluster B Galaxy B Level
Agent Memory Manipulation - ATR-2026-00075 (2983ea65-2ace-56b0-b1c1-11ac28b0525b) Agent Threat Rules LLM Prompt Injection (19cd2d12-66ff-487c-a05c-e058b027efc9) MITRE ATLAS Attack Pattern 1
Agent Memory Manipulation - ATR-2026-00075 (2983ea65-2ace-56b0-b1c1-11ac28b0525b) Agent Threat Rules Stored Data Manipulation - T1565.001 (1cfcb312-b8d7-47a4-b560-4b16cc677292) Attack Pattern 1
Data Manipulation - T1565 (ac9e6b22-11bf-45d7-9181-c1cb08360931) Attack Pattern Stored Data Manipulation - T1565.001 (1cfcb312-b8d7-47a4-b560-4b16cc677292) Attack Pattern 2