← 研踪 PaperTrailarXiv:2603.00902 · v3
Clawdrain: Exploiting Tool-Calling Chains for Stealthy Token Exhaustion in OpenClaw Agents
首次提交 2026-03-01 11:49
版本更新 2026-11-16 06:47
雷达观测 2026-11-17 17:12
摘要
Tool-calling LLM agents rely on third-party skills to extend their capabilities, but this openness exposes a supply-chain attack surface in which malicious skill instructions can manipulate agent execution across multiple turns. In this work, we present Clawdrain, a Trojanized OpenClaw skill that induces resource exhaustion through a Segmented Verification Process (SVP). SVP combines stateful progression, repair-driven repetition, and termination to create costly tool-calling chains while preserving a legitimate final payload. We evaluate Clawdrain in a production-like OpenClaw deployment with real API billing across Gemini 2.5 Pro, DeepSeek V4 Flash, and Claude Sonnet 4.6. Clawdrain achieves 2-16 times token amplification with plain SVP, while the three models exhibit distinct behaviors, including autoregressive compliance, proactive scripting, and extended trial-and-error reasoning. Our findings reveal that agent resource exhaustion arises not only from output generation, but also from repeated context re-ingestion, tool coordination, reasoning, and recovery, motivating system-level controls over cumulative skill behavior.
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v3 · 2026-11-16 06:47 · 最新观测版本
Clawdrain: Exploiting Tool-Calling Chains for Stealthy Token Exhaustion in OpenClaw Agents
Tool-calling LLM agents rely on third-party skills to extend their capabilities, but this openness exposes a supply-chain attack surface in which malicious skill instructions can manipulate agent execution across multiple turns. In this work, we present Clawdrain, a Trojanized OpenClaw skill that induces resource exhaustion through a Segmented Verification Process (SVP). SVP combines stateful progression, repair-driven repetition, and termination to create costly tool-calling chains while preserving a legitimate final payload. We evaluate Clawdrain in a production-like OpenClaw deployment with real API billing across Gemini 2.5 Pro, DeepSeek V4 Flash, and Claude Sonnet 4.6. Clawdrain achieves 2-16 times token amplification with plain SVP, while the three models exhibit distinct behaviors, including autoregressive compliance, proactive scripting, and extended trial-and-error reasoning. Our findings reveal that agent resource exhaustion arises not only from output generation, but also from repeated context re-ingestion, tool coordination, reasoning, and recovery, motivating system-level controls over cumulative skill behavior.
查看此版本解读 →查看此版本原文 ↗