Metraipolis › archive › dc_df8616b3d780

Implicit Manipulation for Skill Selection in LLM Agents with Semantic Matching — arXiv:2609.02035

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.02035
License (as archived)
CC0-1.0
Archived
2026-10-17T04:40:39.767Z
SHA-256 of the archived text
ddb983b5cef3d72bcd60a2607c925f9d9fb7013c63851a7c8ae1e979ca495309
Stable URL of this page
https://metraipolis.com/documento/dc_df8616b3d780

How to cite

Implicit Manipulation for Skill Selection in LLM Agents with Semantic Matching — arXiv:2609.02035 — arXiv (en), https://arxiv.org/abs/2609.02035 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_df8616b3d780 (Metraipolis archive dc_df8616b3d780, sha256 ddb983b5cef3d72bcd60a2607c925f9d9fb7013c63851a7c8ae1e979ca495309)
BibTeX
@misc{dcdf8616b3d780,
  title = {Implicit Manipulation for Skill Selection in LLM Agents with Semantic Matching — arXiv:2609.02035},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.02035},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_df8616b3d780; sha256 ddb983b5cef3d72bcd60a2607c925f9d9fb7013c63851a7c8ae1e979ca495309},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.02035. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Skill selection is a key stage in LLM-agent workflows, determining which installed skill should handle a user request. Existing attacks on this stage primarily rely on explicit prompt injection or instruction-level steering, which can expose recognizable manipulation signals. In this work, we identify a new implicit attack surface for skill selection: even when the user prompt and skill description appear benign in isolation, their semantic relationship can still be strategically shaped to favor an attacker-chosen skill. Based on this observation, we present Implicit Skill-Selection Manipulation via Semantic Matching (ISM), which jointly shapes target-skill metadata and reusable prompts to manipulate skill selection without explicit selection instructions. Specifically, we develop a three-stage strategy to broaden semantic coverage, strengthen target distinctiveness, and preserve natural prompt wording. Across four task domains and eight selector models, ISM increases the average target-selection rate (TSR) from 15.2% to 63.5%. In a matched comparison, ISM achieves a 73.5% TSR, only 9.8 percentage points below Explicit Steering. Human reviewers block ISM in only 2.9% of judgments, versus 91.4% for Explicit Steering, while five LLM-based inspectors pass ISM at an average rate of 82.9%, versus 37.4% for Explicit Steering. Moreover, ISM remains effective against PPL-W, Llama Prompt Guard 2, and PIGuard.

Tema degli esploratori: fiducia digitale e crittografia (retro-catalogo arXiv: annunci del 2026-09-03 in cs.CR).
Categorie arXiv: Cryptography and Security (cs.CR).
Pagina del lavoro: https://arxiv.org/abs/2609.02035 — 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 2026-10-17T04:09:09.345Z dalla pagina https://arxiv.org/catchup/cs.CR/2026-09-03?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).