Metraipolis › archive › dc_d88ea3c183e9
CliffCompaction: Cost-Efficient Compaction for Long-Horizon Coding Agents — arXiv:2609.26779
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.26779
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T13:45:55.233Z
- SHA-256 of the archived text
c724e4c3ec103e86e05e349942cd96501ff7c66f5d47da39aae448d5926db795- Stable URL of this page
https://metraipolis.com/documento/dc_d88ea3c183e9
How to cite
CliffCompaction: Cost-Efficient Compaction for Long-Horizon Coding Agents — arXiv:2609.26779 — arXiv (en), https://arxiv.org/abs/2609.26779 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d88ea3c183e9 (Metraipolis archive dc_d88ea3c183e9, sha256 c724e4c3ec103e86e05e349942cd96501ff7c66f5d47da39aae448d5926db795)
BibTeX
@misc{dcd88ea3c183e9,
title = {CliffCompaction: Cost-Efficient Compaction for Long-Horizon Coding Agents — arXiv:2609.26779},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.26779},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d88ea3c183e9; sha256 c724e4c3ec103e86e05e349942cd96501ff7c66f5d47da39aae448d5926db795},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.26779. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Agents often work on complex problems that require millions of tokens of context, which necessitates compacting across sessions due to limited context windows. We develop CliffCompaction, an autocompaction technique that reduces cost by up to 50% under a bounded context while maintaining or improving performance on Terminal-Bench and achieving new levels of efficiency for test-time scaling and state-of-the-art results on KernelBench. The per-rollout savings of CliffCompaction make the performance--cost trade-off of test-time scaling more efficient, adding over 10 percentage points on Terminal-Bench for less than the cost of two full-context runs. Under parallel test-time scaling, CliffCompaction lets Kimi K2.6 match Opus 4.7, and exceed Opus 4.6 and GPT-5.3 Codex at lower cost. The key to CliffCompaction's effectiveness is that it keeps compacted information faithful by only truncating or dropping content, never rephrasing or rewriting it. We never compact a compaction---each pass operates only on original content, and prior compacted output is discarded, preventing context drift from accumulating. These properties sustain continual learning over sessions exceeding a million tokens: on KernelBench, CliffCompaction reaches CUDA kernel speedups of $2.23\times$ after 200 steps and $3.58\times$ after 400 steps, surpassing specialized search algorithms and trained agents despite being a general-purpose compaction technique. We open-source a scaffold-agnostic API-proxy implementation of CliffCompaction usable with Claude Code, Codex and other harnesses. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-23 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI) ; Machine Learning (cs.LG); Software Engineering (cs.SE). Pagina del lavoro: https://arxiv.org/abs/2609.26779 — 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 2026-10-08T13:18:38.181Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-23?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).