Metraipolis › archive › dc_e99e23d664e2

EvoUndo: Recoverability-Constrained Self-Evolution for LLM Agent Harnesses — arXiv:2608.28363

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.28363
License (as archived)
CC0-1.0
Archived
2026-10-17T10:47:37.208Z
SHA-256 of the archived text
5f7955d9f6e75d901f3a0c27c741a368f73cc653d96ec3405f5bfd2335744744
Stable URL of this page
https://metraipolis.com/documento/dc_e99e23d664e2

How to cite

EvoUndo: Recoverability-Constrained Self-Evolution for LLM Agent Harnesses — arXiv:2608.28363 — arXiv (en), https://arxiv.org/abs/2608.28363 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e99e23d664e2 (Metraipolis archive dc_e99e23d664e2, sha256 5f7955d9f6e75d901f3a0c27c741a368f73cc653d96ec3405f5bfd2335744744)
BibTeX
@misc{dce99e23d664e2,
  title = {EvoUndo: Recoverability-Constrained Self-Evolution for LLM Agent Harnesses — arXiv:2608.28363},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.28363},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e99e23d664e2; sha256 5f7955d9f6e75d901f3a0c27c741a368f73cc653d96ec3405f5bfd2335744744},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.28363. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

LLM agents increasingly modify their own prompts, tools, middleware, resources, and execution harnesses at runtime. Such self-evolution can improve capability, but a successful mutation may leave persistent effects that cannot be safely reversed in states different from the one in which it was created. We introduce EvoUndo, a framework for representing, synthesizing, diagnosing, and independently verifying recoverability of model-generated self-modifications across counterfactual states. Across 600 unseen one-shot self-evolution tasks, we identify 197 capability-improving mutations that fail recoverability verification. Under the original recovery representation, conventional repair strategies recover 0/197 of these natural failures. Deterministic oracle analysis recovers 48/197 under the original recovery language L0, while the extended recovery calculus increases empirical oracle recovery to 191/197. A protocol-locked 2x2 grounding-by-expressivity intervention then separates two bottlenecks: exact state-address grounding increases successful recovery from 0/48 to 38/48 (79.2%) when the original language is sufficient, while extending the recovery language enables recovery on 142/143 (99.3%) failures in the oracle-defined S1 stratum. On the primary gpt-oss-120b backbone, adding exact-address diagnostics to the richer language reduces recovery to 133/143 (93.0%); a Qwen3.8-27B replication preserves the grounding and expressivity effects but not this negative interaction, indicating that the latter is model-dependent. These results indicate that reliable agent self-evolution requires co-designing verification, state grounding, witness semantics, and recovery-language expressivity rather than relying on iterative prompting alone.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-31 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.28363 — 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 2026-10-17T10:33:42.422Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-31?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).