Metraipolis › archive › dc_07c7d7968db5
Poison Valleys: Diagnosing Path-Dependent Failure in LLM Code Repair — arXiv:2610.21514
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.21514
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T09:24:09.745Z
- SHA-256 of the archived text
b0aaf125159609e08b699a6a68682a760d35a9946d613f44fd4969521a14bd3e- Stable URL of this page
https://metraipolis.com/documento/dc_07c7d7968db5
How to cite
Poison Valleys: Diagnosing Path-Dependent Failure in LLM Code Repair — arXiv:2610.21514 — arXiv (en), https://arxiv.org/abs/2610.21514 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_07c7d7968db5 (Metraipolis archive dc_07c7d7968db5, sha256 b0aaf125159609e08b699a6a68682a760d35a9946d613f44fd4969521a14bd3e)
BibTeX
@misc{dc07c7d7968db5,
title = {Poison Valleys: Diagnosing Path-Dependent Failure in LLM Code Repair — arXiv:2610.21514},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.21514},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_07c7d7968db5; sha256 b0aaf125159609e08b699a6a68682a760d35a9946d613f44fd4969521a14bd3e},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.21514. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
When code-repair agents fail, final tests reveal that a patch is wrong but not which earlier commitment caused it or which software obligation failed. Poison Valleys traces this path in 1,844 model requests on two synthetic Python cases: PersistQueue adds atomic batch claiming to a disk-backed queue, while Dijkstra adds source-sensitive filtering to shortest-path search. We define plausible but incorrect software mechanisms, place each in a model's hypothesis, implementation plan, or both, and evaluate the generated code with an Oracle and targeted diagnostic coordinates. Moving the misconception from hypothesis to plan increased failure by 32.2% in PersistQueue and 28.9% in Dijkstra after matching the experimental mix. Different misconceptions produced distinct, model-dependent patterns of violated obligations, although several could converge on the same behavior. This creates a measured connection from upstream misconception to code mechanism and failure form. Code, data, and demo: this https URL . Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Software Engineering (cs.SE) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.21514 — 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 2026-10-16T08:44:17.960Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).