---
title: "StagedWorkspace: A Versioned Workspace for Knowledge-Work Agents"
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","Workspace Versioning","Knowledge-Work Agents"]
---

# StagedWorkspace: A Versioned Workspace for Knowledge-Work Agents

## AI commentary

> This paper reminds us that version ambiguity in office-file handling is often a hidden failure source for agents. Teams building multi-step workflow agents should pay special attention, since ensuring that what an agent sees, edits, and submits stays consistent is a fundamental requirement for reliable automation.

## At a glance

StagedWorkspace proposes a versioned workspace for knowledge-work agents, binding parsed views, review diffs, and submitted artifacts to content hashes of native files, addressing consistency issues when agents search, edit, review, and submit different versions of the same work product. In fixed-harness ablations on OfficeQA Pro and APEX-Agents, dual parsed/native access yields the highest point estimates for every tested model. Adding visible diffs also improves scores on file-editing tasks. The paper argues that workspace state should be treated as an experimental variable in knowledge-work agents and motivates benchmarks that score evidence, staged edits, and submitted artifacts as explicit state transitions.

## Key points

- StagedWorkspace binds parsed records and review diffs to content hashes of native files, forming a versioned workspace.
- Dual parsed/native access outperforms single-view access by 4.7-12.1 points across most models on OfficeQA Pro and APEX-Agents.
- SW-AGENT scores 63.9% with Gemini 3.1 Pro on OfficeQA and 42.1 with GPT-5.4 Nano on APEX, substantially above published same-model results.
- The paper identifies workspace state as a key experimental variable and calls for benchmarks that model state transitions explicitly.

## Full text

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.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_6c58eafaeee83d3433ec6763