PaperDebugger: A Plugin-Based Multi-Agent System for In-Editor Academic Writing, Review, and Editing
Junyi Hou, Andre Lin Huikai, Nuo Chen, Yiwei Gong, Bingsheng He
2025-12-05
Summary
This paper introduces PaperDebugger, a new tool designed to help with academic writing directly inside programs like Overleaf, which is used for writing with LaTeX.
What's the problem?
Currently, tools that use large language models to assist with writing are separate from the actual writing editor. This separation limits their ability to understand the full context of a document, like its structure, past changes, and current state, making it hard for them to offer truly helpful, intelligent assistance within the writing process itself.
What's the solution?
The researchers created PaperDebugger as a plugin that runs *inside* the editor. It uses multiple 'agents' powered by large language models and can connect to other tools for things like searching for research papers and checking references. They overcame technical hurdles like keeping the editor and the tool synchronized, managing different versions of the document, and ensuring everything is secure, using a special system for communication between the tool and the editor.
Why it matters?
This work is important because it shows how to build AI writing assistants that are deeply integrated into the writing environment, rather than being separate programs. This allows for a more seamless and powerful writing experience, and early testing shows people are actually using and finding it helpful, paving the way for more advanced AI-powered writing tools in the future.
Abstract
Large language models are increasingly embedded into academic writing workflows, yet existing assistants remain external to the editor, preventing deep interaction with document state, structure, and revision history. This separation makes it impossible to support agentic, context-aware operations directly within LaTeX editors such as Overleaf. We present PaperDebugger, an in-editor, multi-agent, and plugin-based academic writing assistant that brings LLM-driven reasoning directly into the writing environment. Enabling such in-editor interaction is technically non-trivial: it requires reliable bidirectional synchronization with the editor, fine-grained version control and patching, secure state management, multi-agent scheduling, and extensible communication with external tools. PaperDebugger addresses these challenges through a Chrome-approved extension, a Kubernetes-native orchestration layer, and a Model Context Protocol (MCP) toolchain that integrates literature search, reference lookup, document scoring, and revision pipelines. Our demo showcases a fully integrated workflow, including localized edits, structured reviews, parallel agent execution, and diff-based updates, encapsulated within a minimal-intrusion user interface (UI). Early aggregated analytics demonstrate active user engagement and validate the practicality of an editor-native, agentic writing assistant. More details about this demo and video could be found at https://github.com/PaperDebugger/PaperDebugger.