ScholarScribe: A Privacy-First Local LLM Writing Companion for Researchers

Abstract

ScholarScribe is a desktop application that supports researchers in the preparation of scholarly manuscripts through a suite of local, privacy-preserving tools. The application runs entirely on the user's device, requires no cloud-based application programming interfaces (APIs), collects no telemetry, and transmits no user-generated content to any external host. Drawing on contemporary research into the limitations of AI-generated text detectors (Liang et al., 2023; Weber-Wulff et al., 2023), ScholarScribe is explicitly designed not to facilitate the evasion of such detectors. Instead, it offers transparent, auditable tools for text cleaning, citation validation, document structure analysis, style analysis, AI-use disclosure generation, and structured abstract drafting—functions that align with established norms of research integrity. The application is built on the Tauri 2 framework (Rust backend, Svelte frontend) and uses Ollama as its local large language model (LLM) runtime, enabling the use of open-weight models such as Google's Gemma 3, Alibaba's Qwen 3, Microsoft's Phi-4, DeepSeek R1, and Meta's Llama 3.3 without reliance on proprietary inference services.