Metraipolis › archive › dc_f65e577893d7

Adaptive Consistency Graph for Long-Horizon Agents — arXiv:2609.32754

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.32754
License (as archived): CC0-1.0
Archived: 2026-10-07T20:56:02.206Z
SHA-256 of the archived text: 9063b54144798de1b0350c62410fb9483995ee9b278c840a83da6a87fced5f2a
Stable URL of this page: https://metraipolis.com/documento/dc_f65e577893d7

How to cite

Adaptive Consistency Graph for Long-Horizon Agents — arXiv:2609.32754 — arXiv (en), https://arxiv.org/abs/2609.32754 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f65e577893d7 (Metraipolis archive dc_f65e577893d7, sha256 9063b54144798de1b0350c62410fb9483995ee9b278c840a83da6a87fced5f2a)

BibTeX

@misc{dcf65e577893d7,
  title = {Adaptive Consistency Graph for Long-Horizon Agents — arXiv:2609.32754},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.32754},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f65e577893d7; sha256 9063b54144798de1b0350c62410fb9483995ee9b278c840a83da6a87fced5f2a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32754. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language model agents can often make reasonable local decisions on short tasks, yet their performance degrades when success requires long sequences of dependent actions and tool calls. During execution, task requirements, historical evidence, and the current execution state may gradually become disconnected, so later decisions can drift from the original objective. We study this problem by introducing the Adaptive Consistency Graph (ACG) for long-horizon execution. ACG incrementally organizes execution evidence and its provenance in a persistent graph, then constructs a temporary requirement-centered view for each decision under a bounded context budget. Rather than replacing the base agent's planner or tool executor, ACG provides a structured and traceable context view for each decision. In the matched evaluation, ACG improves GPT-5.6-luna's average success from 44.5\% with ReAct to 50.2\%, with the largest gain on BrowseComp-Plus (73.5\% versus 62.4\%). We further analyze trajectory structure and inference cost to characterize this improvement. Our code is available at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.32754 — 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 2026-10-07T20:29:00.958Z 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).