Metraipolis › archive › dc_9f7319702afe

How Much Skill Orchestration Can LLM Agents Compile Away? Execution-Grounded Specialization for Token-Efficient Skill Use — arXiv:2611.10022

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.10022
License (as archived)
CC0-1.0
Archived
2026-11-10T18:42:08.522Z
SHA-256 of the archived text
3b42331569be0cc5e5469142f238eadd20d843d910da1a6734c671084d4a10f8
Stable URL of this page
https://metraipolis.com/documento/dc_9f7319702afe

How to cite

How Much Skill Orchestration Can LLM Agents Compile Away? Execution-Grounded Specialization for Token-Efficient Skill Use — arXiv:2611.10022 — arXiv (en), https://arxiv.org/abs/2611.10022 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_9f7319702afe (Metraipolis archive dc_9f7319702afe, sha256 3b42331569be0cc5e5469142f238eadd20d843d910da1a6734c671084d4a10f8)
BibTeX
@misc{dc9f7319702afe,
  title = {How Much Skill Orchestration Can LLM Agents Compile Away? Execution-Grounded Specialization for Token-Efficient Skill Use — arXiv:2611.10022},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.10022},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_9f7319702afe; sha256 3b42331569be0cc5e5469142f238eadd20d843d910da1a6734c671084d4a10f8},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.10022. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Reusable skills guide large language model (LLM) agents through recurring requests, but each invocation still requires the model to interpret the skill and reconstruct its orchestration. Can agents automatically compile this orchestration into an executable fast path without human intervention? We first study zero-shot skill specialization, which compiles without execution evidence, and find that despite the skill's procedural guidance, incorrect assumptions about runtime bindings can make generated plans brittle and recovery costly. This motivates SkillCache, a one-trace skill specialization method that compiles a completed invocation into a parameterized capsule of programs and natural-language steps, replays it before reuse, and drives it with a plan executor that returns control to the model only when needed, keeping the source skill for recovery. With the Claude Code harness, warm execution of the same request reduces tokens on 87 SkillsBench tasks by 81% with Claude Sonnet 4.6 and by 74% with Claude Opus 4.8 at comparable reward, and capsules stay effective under controlled request changes. Capsules also transfer across models, raising three open-weight models' mean reward by 12-25 percentage points while reducing their tokens by 72-81%, suggesting that execution-grounded compilation is a promising approach to efficient repeated skill use and cross-model transfer.

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); Computation and Language (cs.CL); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2611.10022 — 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 2026-11-10T18:37:37.566Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).