Direct workspace tabs help tracking complex knowledge tasks in lLM agents
RunningTab: Direct Workspace Interaction with Environment-Side Tabs
Artificial IntelligenceComputation and Language
Summary
Handling many files for a single task is tricky, especially when an AI assistant reads some files but forgets others it should check. The authors introduced RunningTab, a system that keeps track of what an AI has read, which files it should still consider, and what requirements remain unmet, all stored alongside the task environment rather than inside the AI model. This helps the AI better organize information and avoid missing important parts before delivering results.
What this means in practice
- •For software development teams: Keep track of project tasks by recording files read and still to be reviewed to ensure comprehensive code or documentation updates.
- •For legal document processors: Manage and monitor review progress across many legal files, reducing omissions in case preparation or contract analysis.
Authors
Jinheon Baek, Soyeong Jeong, Yumin Choi, Dongsu Han, Sung Ju Hwang
Abstract
Much knowledge work produces new deliverables from files a workspace already holds, and LLM agents are beginning to take such work over. Through direct corpus interaction, an agent can search and read any of those files from a terminal with no indexing, and producing a deliverable from many of them in this way is what we call direct workspace interaction (DWI). Reaching the files, however, is only half the task: nothing keeps track of what the task asks for, what has been read, and what was listed but never opened, all of which slip through the context window without leaving a trace, so an agent may extract a figure and still deliver a report without it. To address this, we present RunningTab, a framework that equips direct workspace interaction with an environment-side tab: a per-task record of what the task still owes, kept by the environment alongside the agent. Specifically, the agent adds its requirements, while the environment records every file read as an excerpt with its provenance and every listed but unopened file as a candidate; the agent can then see each requirement beside its best-matching excerpts and top unopened candidates, resolve it against matching content or set it aside with a reason, and, should it try to finish with requirements still open, receive them in a finish check. We validate RunningTab on three benchmarks with three LLMs, where it consistently outperforms plain DWI and baselines that keep the record in the model, while its tab usually holds the values a deliverable needs once seen.