xiaoyaosearch

Overview of xiaoyaosearch

What it is

XiaoyaoSearch is a cross-platform local desktop application (Windows/MacOS/Linux) designed for knowledge workers, content creators, and technical developers. It uses integrated AI models to support voice input (within 30 seconds), text input, and image input, converting user queries into semantic searches for deep local file retrieval. It was fully implemented via Vibe Coding and provides all source code and development documentation for learning and exchange.

Key features

• Multimodal input: Supports voice recording, text input, and image upload
• Deep search: Searches content and filenames of videos (mp4, avi), audio (mp3, wav), documents (txt, markdown, office, pdf)
• AI enhancement: Integrates BGE-M3, FasterWhisper, CN-CLIP, OLLAMA models; supports cloud large models (OpenAI/DeepSeek/Aliyun compatible APIs) and cloud embedding models (v1.6.0); includes a professional terminology library that improves search recall by 60% in specialized domains (v1.9.0)
• High performance: Hybrid retrieval architecture based on Faiss vector search and Whoosh full-text search
• Privacy controllable: Runs locally with data not uploaded by default; optional cloud API for performance-privacy trade-off
• Notion warm and bright design: Provides a complete design system with a 67% visual comfort improvement (v2.0.0); zero-network-dependent system font stack, Lucide Icons, multi-layer shadows
• AI ecosystem integration: MCP server support for connecting with Claude Desktop (v1.4.0); Agent Skills support for standardized tool calls in Claude Code, VS Code, Cursor (v1.5.0)

Use cases

• Searching for local files on Windows, macOS, or Linux using natural language, voice, or images
• Finding specific content within videos, audio files, or documents on a user's machine
• Knowledge workers and content creators needing fast, semantic file search without uploading data to the cloud
• Developers integrating AI-powered search into their workflows or AI assistants via MCP/Agent Skills

Getting started

See the repository README for detailed installation instructions, including the integrated package method (recommended for non-developers on Windows) and developer deployment method (for Windows/macOS/Linux).

License and pricing

Licensed under AGPL-3.0 for open-source use (free, modifications must be open-sourced). A commercial license is available for closed-source commercial use. Pricing for the commercial license is not stated in the repository.

Based on the project's public README and repository metadata.