DCAS: Decoupling CLI Agent Scaffolding to Internalize Planning across Scaffolds
Source: https://arxiv.org/abs/2608.06113
PDF: https://arxiv.org/pdf/2608.06113v1
Content fetched: 2026-08-09T15:34:20.317851+00:00
Grounding: OpenCLI arXiv metadata and abstract; Obsidian evidence: OpenClaw定时任务/论文流水线/2026-08-09-论文流水线.md
Metadata
- Author(s): Kishanthan Thangarajah, Boyuan Chen, Ahmed E. Hassan
- Original date: 2026-08-06
- Platform: arxiv
- AAIF quality score: 5
中文摘要
论文指出 CLI 软件工程 agent 的微调数据高度绑定 OpenHands 脚手架，模型在训练 scaffold 下得分较高，迁移到其他 scaffold 会显著退化。DCAS 作为后端替换拦截层，在不修改 scaffold 的前提下连接任意 CLI scaffold 与后端模型，用于跨 scaffold 评测和规划感知轨迹采集。实验把显式规划与隐式规划结构分开，显示 planning quality 是缓解迁移掉点的高杠杆因素。
English Summary
DCAS is a backend-substitution interception layer for routing API traffic between any CLI scaffold and backend model without modifying the scaffold. The paper argues that fine-tuning on OpenHands-bound trajectories induces scaffold-specific planning behavior, and shows that planning-aware trajectories collected through DCAS improve transfer across non-training scaffolds.
Abstract excerpt
CLI-based software-engineering agents have matured rapidly, yet the open ecosystem has converged on a single training environment: trajectory datasets used to fine-tune open models are collected almost exclusively under OpenHands. Models fine-tuned on this data score well under OpenHands but degrade substantially when deployed under any non-training scaffold. Untrained base models do not show this divergence, indicating the gap is fine-tuning-induced and tied to the conventions of the training scaffold. We argue that a load-bearing scaffold-specific behavior is planning structure, in two senses this paper distinguishes: explicit planning, a pre-execution plan produced as a first-class artifact, and implicit planning, the structural conventions that shape execution throughout the agent loop. Under this hypothesis, closing the gap requires moving planning from a fixed scaffold artifact to a learned model capability. We introduce Decoupling CLI Agent Scaffolding (DCAS), a backend-substitution interception layer that routes API traffic between any CLI scaffold and any backend model without modifying the scaffold, enabling cross-scaffold evaluation and planning-aware trajectory collection.
Why it matters for AAIF
CLI agent 训练不能只记住某个 scaffold 的规划格式，planning 需要变成模型可迁移能力。