SkillSieve: A Hierarchical Triage Framework for Detecting Malicious AI Agent Skills

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hou, Yinghan, Yang, Zongyou, Pang, Zaihu, Ma, Xiujun
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916048871096320
author Hou, Yinghan
Yang, Zongyou
Pang, Zaihu
Ma, Xiujun
author_facet Hou, Yinghan
Yang, Zongyou
Pang, Zaihu
Ma, Xiujun
contents OpenClaw's ClawHub marketplace hosts tens of thousands of community-contributed agent skills (49,592 in our 2026-04-04 snapshot), and recent audits report that 13-26% contain security vulnerabilities. Regex scanners miss obfuscated payloads; formal static analyzers cannot read the natural-language SKILL.md instructions that hide prompt injection and social engineering. Neither approach covers both modalities. SkillSieve is a three-layer detection framework that applies deeper analysis only where needed. Layer 1 runs regex, AST, and metadata checks through a recall-tuned heuristic scorer, filtering 86% of the volume. Layer 2 routes suspicious skills to an LLM, splitting the analysis into four parallel sub-tasks with structured outputs. Layer 3 puts high-risk skills before a jury of three LLMs that vote independently and debate when they disagree. We evaluate on 49,592 real ClawHub skills and adversarial samples across five evasion techniques, running the pipeline on a 440 USD ARM single-board computer. On a 390-skill labeled benchmark, SkillSieve achieves F1 = 0.920 (precision 0.912, recall 0.929) at 0.006 USD per skill. An optional XGBoost fast-path cuts 32% of Layer-2/3 LLM calls with a 1.6-point F1 reduction, while preserving full-pipeline recall (0.929). For cross-ecosystem generalization, we adapt the framework to Feishu/Lark and scan 52 real packages, where Layer 2 corrects Layer 1 false positives from domain-specific idioms, suggesting a low-cost adaptation path to similar enterprise platforms. We deploy SkillSieve as a Feishu chat bot for real-time skill vetting. Code, data, and benchmark are open-sourced.
format Preprint
id arxiv_https___arxiv_org_abs_2604_06550
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SkillSieve: A Hierarchical Triage Framework for Detecting Malicious AI Agent Skills
Hou, Yinghan
Yang, Zongyou
Pang, Zaihu
Ma, Xiujun
Cryptography and Security
Artificial Intelligence
OpenClaw's ClawHub marketplace hosts tens of thousands of community-contributed agent skills (49,592 in our 2026-04-04 snapshot), and recent audits report that 13-26% contain security vulnerabilities. Regex scanners miss obfuscated payloads; formal static analyzers cannot read the natural-language SKILL.md instructions that hide prompt injection and social engineering. Neither approach covers both modalities. SkillSieve is a three-layer detection framework that applies deeper analysis only where needed. Layer 1 runs regex, AST, and metadata checks through a recall-tuned heuristic scorer, filtering 86% of the volume. Layer 2 routes suspicious skills to an LLM, splitting the analysis into four parallel sub-tasks with structured outputs. Layer 3 puts high-risk skills before a jury of three LLMs that vote independently and debate when they disagree. We evaluate on 49,592 real ClawHub skills and adversarial samples across five evasion techniques, running the pipeline on a 440 USD ARM single-board computer. On a 390-skill labeled benchmark, SkillSieve achieves F1 = 0.920 (precision 0.912, recall 0.929) at 0.006 USD per skill. An optional XGBoost fast-path cuts 32% of Layer-2/3 LLM calls with a 1.6-point F1 reduction, while preserving full-pipeline recall (0.929). For cross-ecosystem generalization, we adapt the framework to Feishu/Lark and scan 52 real packages, where Layer 2 corrects Layer 1 false positives from domain-specific idioms, suggesting a low-cost adaptation path to similar enterprise platforms. We deploy SkillSieve as a Feishu chat bot for real-time skill vetting. Code, data, and benchmark are open-sourced.
title SkillSieve: A Hierarchical Triage Framework for Detecting Malicious AI Agent Skills
topic Cryptography and Security
Artificial Intelligence
url https://arxiv.org/abs/2604.06550