---
title: "AgentPoison"
created: 2026-05-10
updated: 2026-05-10
type: concept
tags: [agent, memory, safety, security, stub]
sources: []
---

# AgentPoison

AgentPoison refers to backdoor or poisoning attacks against agent memory, where malicious content is inserted into persistent state and later influences agent decisions.

It is a memory-safety counterpart to prompt injection: the harmful instruction can persist beyond a single context window.

## Related

- [[memeovobench-memory-safety]] — Benchmark coverage for memory safety risks.
- [[agent-safety]] — Umbrella page for agent safety.
- [[prompt-injection]] — Related attack surface for agent inputs.