A local-first desktop app for searching documents, images, and audio across devices using natural language.
Mango Finder is a local-first desktop application that allows users to search their documents, images, and audio files across multiple devices using natural language queries. It solves the problem of finding information based on content and meaning rather than file names or folder structures, making personal and team knowledge bases instantly accessible. The app supports cross-device search on local networks and processes data locally by default to ensure privacy.
Individuals and teams managing personal document libraries, multi-device environments, or internal knowledge bases who need fast, private, and intelligent search across their local files. It's ideal for users with accumulated PDFs, Word documents, Markdown files, images, and audio across NAS, Mac, Linux, and Windows systems.
Developers choose Mango Finder for its strong privacy guarantees with local-first processing, powerful cross-language and cross-device search capabilities, and flexible deployment options including self-hosted AI models. It works with existing file structures without requiring reorganization and offers both semantic and exact keyword matching.
Search your files across your devices with natural language | 使用自然语言跨设备搜索文件的桌面应用
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
All data processing is done on-device by default, with optional self-hosted models via Ollama/vLLM, ensuring no data leaves user control unless explicitly enabled for remote services.
Supports searching across multiple devices on a local network using mDNS and handles over 100 languages seamlessly, allowing queries in one language to find documents in another without configuration.
Integrates with self-hosted AI models for enterprise intranets, keeping data internal, and includes local models for offline use, as detailed in the architecture diagram.
Automatically detects file changes (add/modify/delete) and updates the index, ensuring search results stay current without manual refreshes.
Requires downloading specific model files, installing Rust and Tauri prerequisites, and configuring OS-dependent tools like LLVM 18 on Windows or environment variables on macOS, which can be error-prone for non-developers.
The cross-device feature relies on mDNS and is prone to failures due to network isolation, firewalls, or subnet differences, often requiring manual troubleshooting as admitted in the FAQ.
Built with Tauri, it only supports Windows, macOS, and Linux desktops, with no mobile or web versions, limiting accessibility for on-the-go use.
Users must manually download and update model files (e.g., from Hugging Face), and dependencies like Whisper.cpp require specific compiler versions, risking compatibility issues over time.