What happened
DigUp version 0.6.2 was released through the project’s GitHub releases page. The update is framed as a set of fixes made during the ten days following version 0.6.1, alongside a new category of searchable files: spreadsheets and presentations. The release is for Apple Silicon Macs running macOS 14 or later; users can update through DigUp’s in-app update command or download the DMG.
The most consequential fix addresses battery drain after the first indexing pass. According to the release notes, two behaviors caused the problem. First, a selected folder containing files that are rewritten continuously, such as a database, log directory, or build output, could make DigUp rescan folders every few seconds. Second, scans involving iCloud Drive or Google Drive folders could take minutes. Version 0.6.2 ignores changes that cannot matter to the index and changes the crawl so that it no longer asks the cloud about every individual file.
The release also changes how selected folders are handled. A chosen folder is now read in full regardless of what it contains. Previously, a home folder with Git dotfiles or a repository containing notes and pictures could be categorized as a code project and retain only screenshots. Settings now reports why files cannot be searched: a folder can show a count of unread files and open the reasons, including files inside code projects, cloud-only files, unsupported types, and Photos libraries.
Several indexing and interaction fixes target long-running use. Progress no longer restarts when files arrive during an indexing pass, and resuming immediately after a pause starts indexing rather than waiting for another change. The search panel can open over a full-screen application when a DigUp window is already open or minimized, rather than switching macOS Spaces. Broken PDF metadata no longer crashes indexing; protected M4P music files are skipped; and cache directories marked with CACHEDIR.TAG are excluded from both indexes.
The new format support covers XLSX, PPTX, ODS, and ODP files. DigUp reads spreadsheet text, sheet names, and rows, plus slide text and speaker notes from presentations. It also indexes Typst files as documents. The release notes report a measured 18–20% improvement in keyword search on a 29,000-vector index after the database was memory-mapped. Other changes include better handling of tiny stub files in generic searches, fewer cloud-folder reads during permission checks, and a security change preventing code search from running Git programs configured by a repository.
What the archive adds
This is a new release and new set of operational facts for a product already covered in DigUp is a free Mac menu bar app for semantic search across files, built on EmbeddingGemma 2. The earlier note established DigUp as a small local semantic-search application; version 0.6.2 adds evidence about its file coverage, indexing behavior, and the practical cost of monitoring folders.
Trend signal
No supplied pattern is a genuinely close fit. This release is primarily an incremental product-maintenance update rather than evidence of a broader change in agent architecture, interoperability, or organizational practice.
Why it matters
Local search tools only become dependable parts of a workflow when their background work is predictable. A semantic index that repeatedly rescans changing folders or turns cloud locations into minute-long scans imposes a hidden operating cost: battery use, delayed updates, and uncertainty about whether results reflect the current files. The fixes matter more than their patch-release framing suggests because they address the boundary between a useful personal index and an always-running process that users may disable.
The additional spreadsheet and presentation support also widens the index beyond prose documents and source code. Teams often keep planning, operating metrics, project status, slide notes, and decisions in those formats. For a builder evaluating local search, the important question is not simply whether a tool can embed files, but which useful content it extracts and how visibly it explains what has been omitted.
Limits and uncertainty
The release notes describe the changes but do not provide independent battery benchmarks, end-to-end indexing times, or a comparison with earlier versions under the same workload. The reported 18–20% figure applies to keyword search on one 29,000-vector index and should not be read as a general performance guarantee. The source also does not establish how well extracted spreadsheet rows, slide text, or speaker notes rank in real searches, nor does it describe privacy, storage, or resource behavior beyond the listed fixes.
Skill to build now
Audit any local or agent-facing indexer against a deliberately noisy folder: build output, logs, cloud-synced files, and files that change continuously. Measure whether irrelevant writes trigger full rescans, make exclusions explicit, and show users why a file is unavailable. Then test representative spreadsheets and presentations rather than treating file-extension support as proof that the useful text is actually extracted and retrievable.
Reader takeaway
DigUp 0.6.2 is a practical reliability release: it aims to stop unnecessary rescans and cloud-folder work while bringing spreadsheets and presentations into the searchable corpus. For builders, its clearest lesson is that file coverage and indexing discipline are inseparable parts of a useful local AI search experience.
Open question
Will the battery and cloud-folder fixes hold under large, actively changing workspaces, and will the newly extracted spreadsheet and presentation content produce useful search results in day-to-day retrieval?