Metraipolis › archive › dc_4d389333eebe

Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents — arXiv:2609.26760

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.26760
License (as archived)
CC0-1.0
Archived
2026-10-08T13:45:50.986Z
SHA-256 of the archived text
644d59728886c17cd16ae2667c81723412bd4475aab1046cd2bb86788f529798
Stable URL of this page
https://metraipolis.com/documento/dc_4d389333eebe

How to cite

Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents — arXiv:2609.26760 — arXiv (en), https://arxiv.org/abs/2609.26760 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_4d389333eebe (Metraipolis archive dc_4d389333eebe, sha256 644d59728886c17cd16ae2667c81723412bd4475aab1046cd2bb86788f529798)
BibTeX
@misc{dc4d389333eebe,
  title = {Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents — arXiv:2609.26760},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.26760},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_4d389333eebe; sha256 644d59728886c17cd16ae2667c81723412bd4475aab1046cd2bb86788f529798},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.26760. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language model (LLM) agents often handle streams of related tasks, yet standard harnesses repeatedly ask the model to reconstruct the same control decisions inside each task's context. We study whether task feedback can instead turn recurring control into reusable executable code, while reserving LLM calls for task-specific semantic reasoning. We introduce Growing Harness, a failure-guided training paradigm that learns the agent harness itself from a strategy-free scaffold that exposes fixed model and tool interfaces but encodes no task-solving controller. Function-level execution traces localize each failure to a bounded code surface, an optimizer repairs a window of failures jointly, and a success-first held-out gate rolls back repair sequences that harm prior capability. Accepted edits accumulate in one shared harness, allowing its control structure to emerge from task feedback. Across BrowseComp-Plus and WebArena-Verified with three deployment models from 4B to 120B parameters, Growing Harness achieves the highest mean success in five of six benchmark-model settings and trails the best mean by 0.7 pp. in the sixth. Relative to a Tool-Calling agent, it reduces LLM calls by 76.0-91.8% and deployed-agent inference cost by 74.4-98.6%. On WebArena-Verified, its success remains 44.7-45.3% across model scales, whereas Tool-Calling falls to 6.7% with the 4B model. Ablations show that trace-local edits, joint repair, and gate-based rollback each improve final success. These results show that persistent program growth can move recurring control out of model context and into low-cost code, yielding reusable specialist agents that remain effective with smaller deployment models.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-23 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Software Engineering (cs.SE).
Pagina del lavoro: https://arxiv.org/abs/2609.26760 — 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 2026-10-08T13:18:38.179Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-23?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).