What is DigUp?
DigUp finds files by their contents rather than relying only on filenames and metadata. It uses Google DeepMind's EmbeddingGemma 2 through llama.cpp to search pictures, scanned PDFs, documents, spreadsheets, presentations, audio and video. Natural-language queries can retrieve content across languages, while keyword matching preserves exact identifiers such as invoice numbers and error codes.
The application runs locally on macOS, Windows and Linux, with Windows and Linux distributed as beta releases. Users select the folders to index, and files remain unchanged and are not uploaded. The model requires an initial download; subsequent update checks can be disabled. DigUp requires no account and reports no telemetry.
Optional code search maintains a separate index for selected projects. It finds functions, definitions and notebook cells, then opens results in a supported editor at the relevant line. Code search is disabled by default and excludes ignored files, generated code, vendored code and files that appear to contain credentials.
DigUp screenshots
Explore the interface. Open a screenshot to view it at full size.
Source: official DigUp project page.
Key capabilities
- Multimodal retrieval Searches images, document pages, audio segments and video frames with natural-language descriptions.
- Exact-word matching Matches names, identifiers and document text alongside semantic results.
- Contextual results Previews matching pages and media moments, with direct opening in supported viewers and players.
- Code search Indexes selected repositories separately and opens matching code at the relevant line in supported editors.
- Local indexing Stores vectors and keyword indexes in local SQLite files without uploading indexed content.
- Folder controls Restricts indexing to selected folders and supports additional folder and file-type exclusions.
Where DigUp fits
DigUp fits desktop workflows involving mixed collections of documents, screenshots, photographs and recordings. It can locate a lease clause, recover a screenshot by its subject, or find a discussion within a recording. Developers can enable its separate code index to retrieve implementation details without mixing source results into general file searches.
Before deployment, hardware and operating-system requirements, initial indexing time and index storage needs should be checked. The Mac application requires Apple Silicon. Windows and Linux remain in beta, and the Windows installer is unsigned. Screenshot text recognition and support for legacy Word documents, RAW photos and SVG images are limited to macOS. Files stored only in the cloud are not downloaded for indexing, and opening a result at a specific page or timestamp depends on the selected viewer or player.
Install
brew install --cask arahim3/tap/digupReviews of DigUp
Be the first to share your experience with DigUp.
Write a review
Loading review form…
