ZipSeekA full-disk index and an on-demand search solve related but different problems. An index pays work in advance so repeated queries across a stable collection can return quickly. An on-demand tool starts from the file, folder, or archive you choose and reads that scope when you ask a question.
ZipSeek uses the second model. It does not scan the disk in the background or build a permanent catalog of every document. This is useful for downloaded research packs, external project folders, and archives that you may inspect once and then remove.
What happens during a ZipSeek search
ZipSeek first discovers eligible items in the selected scope. It processes useful files in the current directory before descending into subfolders, so early results do not have to wait for a complete file-tree inventory. Candidate extensions are used for a low-cost first filter, and the selected candidates are then checked against their actual file signatures or container structure.
Documents and archives are parsed only while a user-initiated search is active. Progress shows the current stage and path, and the operation can be cancelled. Selecting the root of a disk or another obviously broad starting point triggers a confirmation because an on-demand search still has to read every eligible item in that scope.
A cache is not a background index
ZipSeek keeps reusable extracted-text and OCR data in the macOS Caches directory. The cache can accelerate a later query against an unchanged document, but it is populated by searches you actually run. It does not discover new files while the app is idle.
The default combined cache limit is 1 GB. Settings shows the current use and offers 256 MB, 512 MB, 1 GB, 2 GB, and 4 GB limits. When the selected limit is reached, least-recently-used data is removed. Entries also expire after 90 days, and changing a source file prevents its old cached content from being reused.
What clearing local data changes
Clear Local Search Data removes extracted-text and OCR caches, recent locations, recent searches, saved searches, and remembered folder access. It does not delete source documents, and it does not change the ZipSeek trial or Pro purchase record.
Reducing the cache limit may make repeated searches do more parsing or OCR, but it does not reduce search coverage. Very large plain-text files are searched in full while deliberately bypassing the reusable text cache, which prevents one file from repeatedly displacing everyday cached documents.
Choose the model that matches the collection
A full-disk index is a strong choice when you search the same large, stable library many times a day and want instant ranking across everything. It can justify the up-front indexing time, background maintenance, and disk footprint.
ZipSeek is a better fit when the scope is explicit, the source may be an archive or removable folder, exact document locations matter, or the collection should not become part of a permanent catalog. Its first pass does real work at query time; the return is a search tied to the files you selected, with local processing and no background full-disk scan.