LogJack: Indirect Prompt Injection Through Cloud Logs Against LLM Debugging Agents
Identifies a critical security vulnerability in LLM-based debugging agents for cloud operations, with practical attack vectors and inadequate existing defenses.
LLM debugging agents that consume cloud logs are vulnerable to indirect prompt injection; LogJack benchmark shows verbatim command execution rates up to 86.2% (Llama 3.3 70B) and remote code execution succeeds on 6/8 models, with cloud guardrails largely failing to detect injections.
LLM debugging agents that consume cloud logs and execute remediation commands are vulnerable to indirect prompt injection through log content. We present LogJack, a benchmark of 42 payloads across 5 cloud log categories, and evaluate 8 foundation models under 3 prompt conditions with 5 independent trials each (n = 160 per model per condition on 32 attack payloads). Under the active condition, verbatim command execution rates range from 0% (Claude Sonnet 4.6) to 86.2% (Llama 3.3 70B). Passive instructions ("do not execute fixes") reduce most models to 0% but Llama still executes at 30.0%. Remote code execution via curl | bash succeeds on 6 of 8 models. Guardrails from AWS, GCP, and Azure largely fail to detect log-embedded injections-Azure Prompt Shield detected only the most obvious payload (1/32), while GCP Model Armor detected none-though they detect identical payloads in isolation. We also observe a novel "sanitize and execute" behavior where a model detects and removes an obvious malicious component but still executes the remaining injected command. Benchmark and harness available at github.com/HarshShah1997/logjack.