Metraipolis › archive › dc_ebdb4c147e10

Skeleton Coding: Program Structure as a Specification for Agents — arXiv:2610.38214

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.38214
License (as archived)
CC0-1.0
Archived
2026-10-31T06:11:23.969Z
SHA-256 of the archived text
82be1e54ae3a0e7063d031f3b15a330093d65f743c5c13a135c35a09d5049a12
Stable URL of this page
https://metraipolis.com/documento/dc_ebdb4c147e10

How to cite

Skeleton Coding: Program Structure as a Specification for Agents — arXiv:2610.38214 — arXiv (en), https://arxiv.org/abs/2610.38214 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ebdb4c147e10 (Metraipolis archive dc_ebdb4c147e10, sha256 82be1e54ae3a0e7063d031f3b15a330093d65f743c5c13a135c35a09d5049a12)
BibTeX
@misc{dcebdb4c147e10,
  title = {Skeleton Coding: Program Structure as a Specification for Agents — arXiv:2610.38214},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.38214},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ebdb4c147e10; sha256 82be1e54ae3a0e7063d031f3b15a330093d65f743c5c13a135c35a09d5049a12},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.38214. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Coding agents are typically provided end-to-end behavioral specifications that the code they produce should satisfy, such as tests and input--output examples. How many such specifications must a developer provide, and which ones, for the agent to produce the developer's intended program? We argue that this cannot be determined a priori, for it depends on the structure of the implementation, which is left to the agent to decide. An agent first commits to a structure---a decomposition into data types, modules, and functions---which fixes a class of candidate programs; behavioral specifications then select a program within that class. A set of tests that suffices for one structure may leave another badly underdetermined, requiring different edge cases or exponentially more examples. Since the structure is chosen by the agent, developers cannot ensure their specification is complete without reverse-engineering it from thousands of lines of generated code. To resolve this, we introduce Skeleton Coding, in which developers jointly specify the structure---in high-level, code-like format---alongside behavioral specifications. Theoretically, we model AI-assisted coding as a control problem, and show that skeletons can reduce exponentially the number of tests developers must supply to steer an agent toward an acceptable program. Empirically, across four models and three benchmarks, the best-performing skeleton specifications reduce mean test error by 48.4% relative to the best end-to-end behavioral specifications. Even when the agent can query a complete reference implementation, adding a skeleton reduces mean test error relatively by up to 27.0%. Skeletons also narrow the average gap between small and large models by about 52.4% while lowering implementation cost by 15--26x, and reduce code erosion by 24.7% on average.

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); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.38214 — 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 2026-10-31T06:07:08.635Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).