Chao Gu, Quanyan Zhu · 2026 · arXiv
DOI: 10.48550/arXiv.2610.31376 · open-access PDF
Dokumen ini mempelajari masalah eksekusi alat yang aman untuk agen model bahasa besar (LLM) di bawah injeksi cepat. Menggunakan jaring Petri, kami memformalisasi hubungan eksekusi antara LLM dan alat-alatnya sebagai masalah kontrol pengawas, di mana tanda-tanda mewakili kemampuan, persetujuan, dan anggaran. Karena transisi hanya menyala jika tanda-tanda yang dibutuhkan hadir, permintaan suntikan yang prasyaratnya tidak terpenuhi atau sudah dihabiskan ditolak terlepas dari formulasi. Menghadapi hasil alat sebagai tidak terkendali, kami menegakkan keamanan oleh pengawas maksimal yang konvensional, dan transfer keamanan ini
Machine translation (Meta M2M100); the English original follows.
This paper studies the problem of safe tool execution for large language model (LLM) agents under prompt injection. Using Petri nets, we formalize the execution harness between the LLM and its tools as a supervisory control problem, in which marks represent capabilities, approvals and budgets. Since a transition fires only if the marks it needs are present, an injected request whose preconditions are unmet or already spent is refused regardless of its wording. Treating tool outcomes as uncontrollable, we enforce safety by the conventional maximally permissive supervisor, and this safety transf
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.31376, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange