MashuPack

MashuPack is a client-side web application designed to prepare local code repositories for use with large language models (LLMs) such as ChatGPT and Claude. It solves the problem of naive context dumps by generating a structured export file with clear file path headers and boundaries, allowing AI tools to grep, search, and navigate the code efficiently, much like a developer would in a terminal. All operations run entirely within the user's browser, with no data uploaded to any server, ensuring complete privacy.

How to use MashuPack

1. Load your project by either dragging and dropping a folder onto the application or clicking "Browse for folder" to use the native file picker. 2. Use the interactive file tree to browse, filter (press `/` to focus search), and select specific files or folders for inclusion. Ticking a folder selects all its contents. 3. Check the project statistics in the right panel, especially the token count estimate (toggle by clicking the Size stat) to ensure it fits within your AI model's context window. 4. Export the selection using one of the options: "Export combined text" to download a `.txt` file for AI upload, "Copy to clipboard" for direct pasting into a chat, or "Save as .txt".

Core Features of MashuPack

• Structured Export Format: Generates a plain text file with a directory tree header and each file wrapped in explicit START/END path markers, enabling reliable navigation by AI models.
• Complete Browser-Based Processing: Uses the File System Access API to read local folders without uploading content. All indexing and computation happens client-side.
• High-Performance Engine: Indexing and selection math run in Rust compiled to WebAssembly, offloaded to a Web Worker to keep the UI smooth even with massive repositories (e.g., 28,000+ files).
• Intelligent File Selection & Filtering: Supports selection by checkbox, file-type pills, or the right-panel file types table. Provides live stats (files, folders, size, token estimate).
• Built-in File Viewer: Preview files with syntax highlighting (powered by CodeMirror) without affecting the export selection.
• Multiple Export Options: Export combined text as a file, copy report to clipboard, or download the project as a ZIP archive.

Use Cases for MashuPack

MashuPack is ideal for developers and teams looking to leverage AI for code analysis, review, documentation, or debugging. It allows you to provide an entire codebase structure to an AI, enabling it to trace dependencies, locate specific functions, and understand project layout without being overwhelmed by a single context dump. It's particularly useful for onboarding new projects, generating summaries, or asking complex, cross-file questions about a repository.

Advantages of MashuPack

The primary advantage is significantly enhancing the utility of AI for code by providing a structured, navigable context rather than a flat text dump. This respects the internal mechanisms of modern LLMs that can parse structured data. Furthermore, it offers robust privacy by keeping all data local, and delivers exceptional performance through its Rust/WASM architecture, making it practical for real-world, large-scale projects. The user interface is intuitive, with powerful selection and filtering tools to refine the exact context sent to the AI.