---
title: "StagedWorkspace：面向知识工作智能体的版本化工作区"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.18050v1"
permalink: "https://feed.astralor.com/items/itm_6c58eafaeee83d3433ec6763"
published: "2026-08-18T17:44:18.000Z"
score: 76
channel: "ai"
tags: ["APEX-Agents","Gemini 3.1 Pro","OfficeQA Pro","StagedWorkspace","工作区版本管理","知识工作智能体"]
---

# StagedWorkspace：面向知识工作智能体的版本化工作区

## AI 点评

> 这篇论文提醒我们，Agent 处理办公文件时的版本混乱往往是隐性失败源。对正在构建多步骤工作流 Agent 的团队尤其值得关注，因为“看到什么、改了什么、交了什么都对得上”是可靠自动化的基础前提。

## 速览

StagedWorkspace提出了一种为知识工作智能体设计的版本化工作区，将解析视图、审查差异和最终产物统一绑定到原生文件的内容哈希上，解决智能体搜索、编辑、审查和提交时可能面对不同版本一致性问题。在OfficeQA Pro和APEX-Agents上的固定框架消融实验中，双解析/原生访问相比单一视图显著提升成绩；加入差异可见性后，文件编辑任务得分更高。论文认为工作区状态应作为知识工作智能体的实验变量，并呼吁基准测试将证据、分阶段编辑与提交产物视为显式状态转换。

## 关键点

- StagedWorkspace 将解析记录与审查差异绑定到原生文件的内容哈希，形成版本化工作区。
- 在 OfficeQA Pro 和 APEX-Agents 上，双解析/原生访问比单一视图在多数模型上提升 4.7-12.1 个点。
- SW-AGENT 搭配 Gemini 3.1 Pro 在 OfficeQA 达到 63.9%，搭配 GPT-5.4 Nano 在 APEX 达到 42.1%，显著超过同模型已发表成绩。
- 研究发现工作区状态本身是知识工作智能体的关键实验变量，基准应显式建模状态转换。

## 全文

AI agents increasingly perform knowledge work (i.e., produce and modify persistent digital artifacts such as code repositories, documents, spreadsheets, slides, reports), yet the parsed views they search, the native files they edit, the changes they review, and the artifacts they submit can refer to different versions of the same work product. We formulate this as a workspace-state contract: every view should be explicitly tied to a version of the evolving workspace state. Coding agents partly address this need through repository contracts for search, diffs, and tests, whereas an analogous contract is less explicit for PDFs, spreadsheets, slides, notebooks, and mixed-format project folders. We propose StagedWorkspace, a versioned workspace for knowledge-work agents. The workspace binds parsed records and review diffs to content hashes of the native files as they change. In fixed-harness ablations on OfficeQA Pro and APEX-Agents, dual parsed/native access has the highest point estimate for every tested model; relative to the more limiting single view, it improves OfficeQA Pass@1 by 8.3-12.1 points and APEX mean rubric score by 4.7-9.2 points. SW-AGENT scores 63.9% with Gemini 3.1 Pro on OfficeQA and 42.1 with GPT-5.4 Nano on APEX, compared with published same-model scores of 29.3% and 25.5, respectively. A paired review-axis ablation on 57 file-editing tasks further finds higher observed scores when diffs are visible. These results identify workspace state as an experimental variable in knowledge-work agents and motivate benchmarks that score evidence, staged edits, and submitted artifacts as explicit state transitions.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_6c58eafaeee83d3433ec6763