Metraipolis › archive › dc_2d5607edc0bf
LatentTriad: Pre-Action Risk Representations for Malicious Agent Skill Detection — arXiv:2611.05088
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.05088
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-04T12:28:41.687Z
- SHA-256 of the archived text
2a967aa796fd605cc6eeb25650691903e67cd99c210deb24f6359d469248273d- Stable URL of this page
https://metraipolis.com/documento/dc_2d5607edc0bf
How to cite
LatentTriad: Pre-Action Risk Representations for Malicious Agent Skill Detection — arXiv:2611.05088 — arXiv (en), https://arxiv.org/abs/2611.05088 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2d5607edc0bf (Metraipolis archive dc_2d5607edc0bf, sha256 2a967aa796fd605cc6eeb25650691903e67cd99c210deb24f6359d469248273d)
BibTeX
@misc{dc2d5607edc0bf,
title = {LatentTriad: Pre-Action Risk Representations for Malicious Agent Skill Detection — arXiv:2611.05088},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.05088},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2d5607edc0bf; sha256 2a967aa796fd605cc6eeb25650691903e67cd99c210deb24f6359d469248273d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.05088. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Large language model agents are increasingly extended with reusable skills, but untrusted skill content can induce operations outside the user's task. Detecting this contextual risk requires more than suspicious-pattern matching, while dynamic analysis incurs the cost of skill execution. We find that hidden states formed during prefill contain readable information about task-operation consistency, including across held-out expression forms in controlled evaluations. Matched Operation-state interventions also affect later Boundary states and shift fixed risk readouts toward the donor label. Based on these findings, we introduce LatentTriad, which combines contextual Boundary features, group-adjusted Operation directions, and departures from benign state relationships. All branches share one prefill: detection generates no output tokens, invokes no tools, and executes no skill code. Across 5,830 skill packages from five sources and seven backbones, LatentTriad achieves 99.0-99.2% pooled F1 and 93.0-94.6% source-macro F1. A separately measured compact implementation completes prefill and detection in a median of 259-284 ms. These results connect pre-action risk readability and functional influence to execution-free skill screening. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2611.05088 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-11-04T12:26:47.187Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).