Metraipolis › archive › dc_e694b585d56f
How Do Agentic LLMs Decide to Call Tools? A Tool-Call Vector Shaped by Suppression — arXiv:2610.09624
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.09624
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T11:09:27.662Z
- SHA-256 of the archived text
d8556f1cfeb314b91cfbc869724b3af2372919d94986c9fe9ad5860b78786ef7- Stable URL of this page
https://metraipolis.com/documento/dc_e694b585d56f
How to cite
How Do Agentic LLMs Decide to Call Tools? A Tool-Call Vector Shaped by Suppression — arXiv:2610.09624 — arXiv (en), https://arxiv.org/abs/2610.09624 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e694b585d56f (Metraipolis archive dc_e694b585d56f, sha256 d8556f1cfeb314b91cfbc869724b3af2372919d94986c9fe9ad5860b78786ef7)
BibTeX
@misc{dce694b585d56f,
title = {How Do Agentic LLMs Decide to Call Tools? A Tool-Call Vector Shaped by Suppression — arXiv:2610.09624},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.09624},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e694b585d56f; sha256 d8556f1cfeb314b91cfbc869724b3af2372919d94986c9fe9ad5860b78786ef7},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.09624. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Tool calling, invoking external tools on demand, is central to agentic LLMs, yet the mechanism that decides whether a model calls a tool or responds directly remains poorly understood. Agentic prompts are long and heavily scaffolded, combining role instructions, tool schemas, format templates, and the user's request across hundreds of tokens, creating a noisy, highly entangled context in which no single controllable variable for mechanistic analysis is obvious. To obtain such a variable, we propose a method that converts complex agentic prompts into minimal contrastive pairs in which a single request verb determines the tool-call decision: replacing an execution-verb (e.g., \textit{write}) with an analysis-verb (e.g., \textit{discuss}) reliably flips the decision, suggesting it is mediated by a compact internal state. We construct 500 such paired prompts across Python, Java, and C++ (300 for mechanistic analysis, 200 held out for evaluation). We trace the decision to a vector, $\mu_\Delta$, that is both causally necessary and sufficient and generalizes beyond the discovery prompts to native multi-turn $\tau^2$-Bench trajectories and verb-free requests. Behavioral ablations show that the scaffold establishes a tool-call prior; Transcoder decomposition then reveals that analysis verbs suppress this prior through features signaling that tool use is unnecessary, whereas execution verbs largely leave it intact. Downstream scaffold-reading attention heads and MLP features read out the resulting state, and the same mechanism recurs across seven models from the Qwen, Mistral, and Granite families. Our code is available at this https URL .
Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.09624 — 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 2026-10-08T10:14:05.950Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).