Metraipolis › archive › dc_cdeb33740d5d

Self-Evolving Harness on Multiple Tasks with the Agent as Its Own Optimizer — arXiv:2609.38372

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.38372
License (as archived): CC0-1.0
Archived: 2026-10-07T12:41:45.164Z
SHA-256 of the archived text: b30c3666233c39005a29c4162f92d49ed4166723bc74708d8972b73bfed89438
Stable URL of this page: https://metraipolis.com/documento/dc_cdeb33740d5d

How to cite

Self-Evolving Harness on Multiple Tasks with the Agent as Its Own Optimizer — arXiv:2609.38372 — arXiv (en), https://arxiv.org/abs/2609.38372 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cdeb33740d5d (Metraipolis archive dc_cdeb33740d5d, sha256 b30c3666233c39005a29c4162f92d49ed4166723bc74708d8972b73bfed89438)

BibTeX

@misc{dccdeb33740d5d,
  title = {Self-Evolving Harness on Multiple Tasks with the Agent as Its Own Optimizer — arXiv:2609.38372},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.38372},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cdeb33740d5d; sha256 b30c3666233c39005a29c4162f92d49ed4166723bc74708d8972b73bfed89438},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38372. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

A harness is the code around a language-model agent that organizes prompts, calls tools, manages context, and controls execution. As models grow stronger, recent work has begun to let agents improve their own harnesses, a line of work known as self-evolving harnesses. In most existing methods, a separate proposer running on a human-designed harness modifies the solver's harness, and a separate harness is evolved for each benchmark. Real-world tasks come from many domains, so both the evolution and the evaluation of a harness should cover a diverse range of tasks. We propose a framework close to recursive self-improvement: the same frozen model, on the same version of the harness, first solves tasks as the solver and then, as the proposer, reads the complete run records and directly edits the harness that runs it. Each evolution batch draws tasks from five benchmarks in different domains. To measure generalization, training and held-out tasks are strictly separated, and we additionally evaluate on five out-of-distribution benchmarks never used during evolution. We frame the evolution process as deep-learning training with two stages, multi-task pretraining and continual training. Starting from a 49-line seed harness, the harness obtained at the end of the first stage improves the average score by 4.48 points on the in-distribution benchmarks and by 12.64 points on the out-of-distribution benchmarks, surpassing Codex on the former and matching it on the latter. In the second stage, continued evolution on Claw-Eval, one of the out-of-distribution benchmarks, further raises the score on that benchmark from 66.17 to 68.06, exceeding Codex. We also provide an in-depth analysis of the mechanisms that emerged during evolution, including output truncation, history compaction, and independent review.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.38372 — 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 2026-10-07T12:14:13.615Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).