Repolode
Submit
ARahim3

ARahim3/DigUp

Desktop app that searches your local files by meaning, across photos, PDFs, documents, audio, video and code, running on-device with EmbeddingGemma 2.

Open-source alternative to:SpotlightGoogle Photos

View on GitHub
Stars
4,775
Forks
339
Open issues
103
Last commit
yesterday

What is it?

DigUp lets you search the files on your computer by describing what is in them. You choose folders to index, and it reads pictures, PDFs, documents, audio recordings, videos and, optionally, source code. A query like "a dog on the beach" can match a photo, "where they talk about sleep" can jump to a minute of a podcast, and "the clause about pets in the lease" can open the right PDF page.

It is built on Google DeepMind's EmbeddingGemma 2, a multimodal embedding model that places text, images, audio and video in one shared vector space. The model runs locally through llama.cpp, using the GPU where available (Metal on macOS, Vulkan on Windows and Linux). Exact strings such as invoice numbers or error codes are also matched by keyword and boosted in ranking. Nothing you index leaves your machine; the only network traffic is the one-time model download and a daily update check that sends nothing about your files.

The Mac app is written in Swift with SwiftUI and uses macOS frameworks for file reading and screenshot OCR. The Windows and Linux versions share the same search engine and are built with Tauri, with the first Windows and Linux release marked as beta.

Key features

  • 01Semantic search across images, PDFs, documents, audio and video
  • 02Jumps to the matching PDF page, audio timestamp or video moment
  • 03Optional code search that opens results at the matching line in your editor
  • 04Runs fully on-device with llama.cpp, using GPU acceleration where available
  • 05Exact-match keyword search for invoice numbers, codes and names
  • 06Available for macOS, with Windows and Linux builds in beta

Use cases

  • Finding a specific moment in a long recording or video without scrubbing through it
  • Locating a clause or page in a folder full of PDFs and scanned contracts
  • Searching personal photos by what they show rather than by file name
  • Searching your own codebase by describing what a function does

Topics

#semantic-search#embeddings#file-search#on-device#local-first#multimodal#llama-cpp#macos#embeddinggemma#linux

FAQ

What is DigUp?
Desktop app that searches your local files by meaning, across photos, PDFs, documents, audio, video and code, running on-device with EmbeddingGemma 2. DigUp lets you search the files on your computer by describing what is in them. You choose folders to index, and it reads pictures, PDFs, documents, audio recordings, videos and, optionally, source code. A query like "a dog on the beach" can match a photo, "where they talk about sleep" can jump to a minute of a podcast, and "the clause about pets in the lease" can open the right PDF page.
Is DigUp free to use? What license does it have?
DigUp is open source under the MIT license. Check the license terms before using it in commercial products.
Is DigUp still maintained?
Actively maintained. The last commit to DigUp was on Oct 18, 2026.
What can I use DigUp for?
Finding a specific moment in a long recording or video without scrubbing through it Locating a clause or page in a folder full of PDFs and scanned contracts Searching personal photos by what they show rather than by file name Searching your own codebase by describing what a function does
Where can I get DigUp?
The source code and installation instructions are on GitHub: https://github.com/ARahim3/DigUp

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects