Metraipolis › archive › dc_220d55bbdc20

Scope-Gap Injection: Exploiting Evaluator Channel Blind Spots in the OpenClaw Agent Security Competition — arXiv:2610.34777

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.34777
License (as archived)
CC0-1.0
Archived
2026-10-27T07:58:11.937Z
SHA-256 of the archived text
776fff674e8b46d0a66cebbbfe92e40ec7cbbaeb4318a885070cfe35c7841347
Stable URL of this page
https://metraipolis.com/documento/dc_220d55bbdc20

How to cite

Scope-Gap Injection: Exploiting Evaluator Channel Blind Spots in the OpenClaw Agent Security Competition — arXiv:2610.34777 — arXiv (en), https://arxiv.org/abs/2610.34777 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_220d55bbdc20 (Metraipolis archive dc_220d55bbdc20, sha256 776fff674e8b46d0a66cebbbfe92e40ec7cbbaeb4318a885070cfe35c7841347)
BibTeX
@misc{dc220d55bbdc20,
  title = {Scope-Gap Injection: Exploiting Evaluator Channel Blind Spots in the OpenClaw Agent Security Competition — arXiv:2610.34777},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.34777},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_220d55bbdc20; sha256 776fff674e8b46d0a66cebbbfe92e40ec7cbbaeb4318a885070cfe35c7841347},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.34777. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

This paper introduces Scope-Gap Injection (SGI), an indirect prompt injection method for multi-stage LLM-agent safety pipelines. In the IJCAI 2026 OpenClaw Agent Security Attack Competition, each attack package is judged by a static skill scan (S1), an LLM-judge safety check over the task and skill definition (S2), and a live execution audit (S3). Our observations suggest a scope gap: S2 appears to inspect the task instruction and skill definition without inspecting attachment contents, while the agent reads attachments during S3. SGI keeps the visible task and skill benign, while target identifiers, bypass parameters, and authorization procedures are moved into an attachment written as a business worklist. The S2 judge therefore sees a harmless task, while the agent follows the hidden worklist and triggers the permission bypass. In OpenClaw, SGI bypassed all three stages for 5 of 19 targets. The highest-scoring single submission achieved $F = 0.385$. Across the submission campaign, attachment placement was associated with higher aggregate S2 acceptance than in-skill placement: 78% versus 23%. We close with defenses that align evaluator scope with the channels available to the agent.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Cryptography and Security (cs.CR).
Pagina del lavoro: https://arxiv.org/abs/2610.34777 — 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 2026-10-27T07:55:30.714Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).