Skip to main content

Searching Documents

Fiskl searches your document library two ways: by file name, and by the text inside your files. This guide explains the difference, which file types Fiskl can read, and why a file you expected to find sometimes does not appear.

Before You Begin​

You need the document.read permission to search. Content search works on every plan, with no separate allowance to buy or enable.

The Two Search Modes​

The search box sits at the top of the Documents screen, with a Name and Contents toggle beside it.

  • Name matches against file names. A search for "lease" finds office-lease-2026.pdf.
  • Contents matches against the text inside your files. A search for "lease" finds a scanned contract whose file name is scan_0043.pdf, as long as the word appears in the document.

The two are deliberately separate rather than combined. Content search only reaches files Fiskl was able to read, so mixing the modes would make an empty result ambiguous — you would not know whether nothing matched or nothing was readable.

Results appear as you type. Select the X in the search box, or use the Escape key, to clear the search.

How Results Are Presented​

Searching changes the list in three ways:

  • Results cover every folder in your library, not only the folder you are in
  • Folders disappear from the list, since results span the whole library
  • A Folder column appears, showing where each result lives

A count above the list shows how many results matched. Name results are ordered newest first. Content results are ordered by relevance, so the strongest match appears at the top rather than the most recent file.

Deleted documents never appear in either mode.

What Fiskl Can Read​

Extraction runs after the malware scan clears, so a file is searchable by name immediately and by content shortly afterwards.

File typeHow Fiskl reads it
PDF with a text layerReads the text directly
Scanned or photographed PDFReads it with optical character recognition
JPEG, PNG, TIFF up to 10 MBReads it with optical character recognition
Word, Excel, PowerPointReads the text in the file
Plain text, CSV, Markdown, JSONReads the file directly

Fiskl reads the first 50,000 characters of a file. That covers a long contract comfortably, but text beyond that point in a very large document is not searchable.

What Content Search Cannot Find​

These files upload, store and download normally. Fiskl cannot read them, so only name search finds them:

  • Images in WebP, HEIC, GIF, BMP or SVG format
  • JPEG, PNG or TIFF images larger than 10 MB
  • Audio, video, archives such as .zip, and password-protected files
  • Any other format not listed in the table above
info

A photo taken on an iPhone is usually a HEIC file, which Fiskl cannot read. If you want a photographed document to be searchable, save or export it as a JPEG or PDF before uploading.

Common Issues​

Content search finds nothing, but I can see the words in the file

Three things cause this. First, check the file type against the table above — HEIC images and large photos cannot be read. Second, a file uploaded moments ago may still be extracting; try again shortly. Third, if the text sits beyond the first 50,000 characters, it is outside the searchable range.

Search by name as a fallback, then preview the file to confirm.

Searching by name misses a file I know is there

Name search matches the stored file name, and Fiskl replaces unusual characters with underscores when you upload. A file you uploaded as Rapport café.pdf is stored as Rapport_caf_.pdf, so searching for "café" finds nothing. Search for "rapport" instead.

I cannot narrow results by date or file type

Search has no filters. Results always cover the whole library, ordered newest first by name and by relevance for contents. Use the Folder column to see where each result sits, and open that folder to browse it in context.

A scanned document returns odd words

Optical character recognition reads what it can see. Faint scans, handwriting, unusual fonts and skewed pages all reduce accuracy, and the extracted text can contain mistakes. Re-scanning at a higher quality usually improves results. The file itself is never altered.