Kensaku is an open-source, local-first desktop search engine built with Rust and Tauri. It helps you find files by filename, content, and meaning, combining fuzzy filename search, BM25 full-text retrieval, semantic search, and OCR workflows.
Unlike traditional file search, Kensaku lets you search by what you remember about a file, even when you don't remember its exact name or keywords. It supports document extraction across formats such as PDF, Office documents, EPUB, RTF, code, and logs, alongside OCR-enabled workflows for image-based content.
Kensaku also includes an MCP server, allowing compatible AI agents to query the same local index and retrieve relevant snippets. This makes your existing local files accessible to AI-assisted workflows without requiring a separate cloud-hosted search index.
Built with Rust and Tauri, Kensaku uses a compact, static quantized embedding table of approximately 7.7 MB for semantic search, which doesn't require a GPU for queries. The project is designed around local-first use, with no account required and no cloud-based file search.
Kensaku is released under the MIT license. Explore the source code, try the available releases, report issues, or contribute on GitHub: https://github.com/yjid/kensaku
If you find it useful, please consider starring the repository ⭐ to help more developers discover it.
No comments or reviews, maybe you want to be first?