Metraipolis › archive › dc_fc7271a7f399

Action-Space Shaping for LLM Agents: Measuring and Mitigating Tool-Schema Bias — arXiv:2609.34971

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.34971
License (as archived)
CC0-1.0
Archived
2026-10-07T20:58:24.086Z
SHA-256 of the archived text
036065047283b7de480ff57b17b46c7f768696150160570d2c7dc67a77be3f97
Stable URL of this page
https://metraipolis.com/documento/dc_fc7271a7f399

How to cite

Action-Space Shaping for LLM Agents: Measuring and Mitigating Tool-Schema Bias — arXiv:2609.34971 — arXiv (en), https://arxiv.org/abs/2609.34971 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fc7271a7f399 (Metraipolis archive dc_fc7271a7f399, sha256 036065047283b7de480ff57b17b46c7f768696150160570d2c7dc67a77be3f97)
BibTeX
@misc{dcfc7271a7f399,
  title = {Action-Space Shaping for LLM Agents: Measuring and Mitigating Tool-Schema Bias — arXiv:2609.34971},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.34971},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fc7271a7f399; sha256 036065047283b7de480ff57b17b46c7f768696150160570d2c7dc67a77be3f97},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34971. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large Language Models (LLMs) have shown strong performance on tool-use agentic tasks when given a fixed tool schema. Yet a tool schema is not the action space of an agent; it is merely one interface representation of it. The same executable action can be exposed through many different, functionally equivalent tool definitions, and an agent that has truly learned a task should behave consistently across them. We show that current agents often do not, a phenomenon we term schema bias. To study this systematically, we introduce an executable transformation framework that rewrites a native tool schema using nine operators, including merging and splitting tools, altering how a single tool is expressed, and distributing one action across several dependent calls. The tasks, executable actions, and reachable states remain fixed, so any change in success is attributable to the interface alone. Evaluating eleven LLMs, including two closed models, on up to 32 schema variants, we ask how large schema bias is, how it manifests, whether the difficulty of a schema variant can be predicted without a full evaluation, and whether training removes it. We find that schema bias is substantial even for the newest models: success rates range from complete failure to 97% depending solely on the schema. To reliably estimate schema difficulty, it requires running a small sample of the target queries. Training repairs a schema variant only when that variant appears in the training data.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.34971 — 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 2026-10-07T20:29:01.740Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).