futureuniant/workshadow

WorkShadow is a local-first desktop work journal application written in TypeScript, built for people who want to keep detailed records of their daily work, decisions, and progress without storing that data in the cloud. The README is written primarily in Chinese. The application runs as a desktop program using the Tauri framework, which bundles a web frontend with a lightweight native shell. The core feature is a rich text editor on the right side of the screen, where you write journal entries covering tasks, decisions, problems, and notes. Entries support headings, lists, task checkboxes, code blocks, tables, links, images, and video. On the left side is a tree of folders and entries for organizing your journal. Entries are saved to your local machine, not to any external server. Search comes in two forms. Keyword search matches text across all entries and shows preview snippets. Semantic search is available when you connect your own embedding model in the settings, letting you search by describing what you remember rather than typing exact words. The AI features are handled through a workbench panel. You can select multiple journal entries and ask the app to generate a written summary for reports, emails, or updates. You can also ask questions across your entire journal history, and the app will retrieve relevant entries and synthesize an answer with source references. Both features require you to configure your own language model by supplying an API endpoint and key in the settings. No model is bundled in. The repository offers two versions. The developer build can be compiled from source with Node.js and Rust installed. The installer build is a prebuilt Windows executable that includes performance optimizations, image drag-and-drop, and a local autocomplete feature that learns from your own past entries over time. Both versions are free and store all data locally. Backup and migration work through a custom .ws archive format. The code is released under AGPL-3.0.