POISE: Position-Aware Undetectable Skill Injection on LLM Agents

2026-06-05

AI Summary

Existing skill injection attacks against large language model agents struggle to simultaneously achieve high success rates and strong stealthiness. This work proposes a position-aware, stealthy injection method that embeds malicious payloads into a single seemingly benign skill-body instruction through adversarial instruction compression, and seamlessly integrates it into the task’s preparatory or preceding steps via a context generator. The approach uniquely unifies high attack efficacy with low detectability, achieving an 89.3% success rate on the Skill-Inject benchmark—28.0 percentage points higher than random body injection—while triggering new high-risk alerts in only 5.6% of samples, thereby effectively evading static detection mechanisms.

Abstract

Agent skills provide a lightweight mechanism for extending general-purpose agents, but their open format exposes them to skill-poisoning attacks. A practically dangerous injection must stay invisible: if executing the payload derails the user's legitimate task, the resulting failure signal invites inspection of the skill. We therefore evaluate attacks by Attack Success Rate, which requires the injected payload to execute and the user's task to still pass its verifier in the same trial. Prior skill-poisoning attacks face a reliability-stealth trade-off under this lens: YAML-header injections are reliably loaded but easily inspected, whereas stealthier body injections that place explicit malicious commands in the skill prose are less reliable because out-of-context commands invite the agent's own suspicion. We introduce POISE, a position-aware attack that compresses the trigger into a single, benign-looking body instruction, placing it at a feasible position and using a context-aware generator to blend it with nearby setup or prerequisite steps. On Skill-Inject with codex+gpt-5.2, POISE achieves an 89.3% ASR, 28.0 points above a random-placement body baseline and 2.6 points above a YAML-only baseline, while retaining the stealth advantage of body placement. That stealth is the decisive margin: because legitimate skill bodies naturally require privileged tool operations, LLM scanners are hyper-sensitive, falsely flagging 74.6% of clean skills on average across four judges and both benchmarks. Blending into these false alarms, POISE causes only 5.6% of poisoned variants to gain a new high-risk alert over their clean baselines, rendering current static defenses ineffective.

Problem

• skill poisoning
• undetectable attack
• LLM agents
• stealth injection
• attack success rate

Innovation

• position-aware injection
• skill poisoning
• LLM agent security
• stealthy attack
• context-aware generation