Skip to content
Appendix / File formats and limits
Open app

File formats and limits

Uploading a file, previewing it, and extracting usable evidence are separate steps. A successful upload does not guarantee that the file has been parsed or that every analysis tab is available.

Common document formats

FormatProcessing and preview
PDFDocument parsing and PDF preview
Word: DOC, DOCXDocument parsing and Word preview
Excel: XLS, XLSXSpreadsheet processing and workbook preview, subject to browser limits
CSV, TSVDelimited-text processing and spreadsheet preview
Text and MarkdownText processing; common text extensions have an in-app preview
Email: EML, MSGCommunication-thread parsing; review the extracted communication content
PowerPoint: PPT, PPTXDocument parsing; ordinary uploaded slide files do not have a dedicated presentation preview
RTF, DOTX, WPD, QPWSupported parsing routes, without a dedicated preview for each format
XLSM, XLTX, XLTMWorkbook processing; no dedicated workbook preview for these extensions

A presentation produced by Colabra AI can have a PDF preview and an original PowerPoint download. That delivery behavior is separate from preview support for ordinary uploaded slides.

Images, text, and specialist formats

The parsing routes support common scanned-image formats including PNG, JPEG, TIFF, BMP, GIF, HEIC, and HEIF, plus supported legacy image formats. The browser’s ability to display an image can still vary by encoding and browser.

Text processing also recognizes source code, configuration, HTML, JSON, XML, YAML, calendar files, contact cards, and text subtitle formats through supported extensions or MIME types. Preview coverage is narrower: for example, JSON and HTML have dedicated viewers while many source-code files use text preview.

Audio and video can be previewed when the browser supports their codecs. Do not assume that playback means automatic speech transcription or diligence extraction has completed. Upload a transcript when the spoken content needs to be reviewed as evidence.

Archive formats and specialist application files are not a substitute for supported source documents. Extract the relevant files from an archive or export a reviewable format before uploading. Retain the original separately when it is needed for the record.

Upload limits

RouteLimit
Standard app uploadUses the storage bucket’s configured limit; the default validation is 100 MB (100,000,000 bytes) per file
REST / MCP base64 upload20 MiB (20,971,520 bytes) after decoding
Browser preview loadingStandard file-content preview blocks files above 50 MiB

Workspace limits and permissions can affect which operations are available. Check the error shown for the actual upload rather than assuming the browser and API have identical limits.

Browser preview limits

PreviewBehavior
TextFullscreen text is capped at 500,000 characters; inline preview is shorter
CSV / TSVUp to 2,000 rows fullscreen and 5 rows inline
Excel workbookPreview is blocked above 120 sheets, 150,000 rows, or 3,000,000 total cells
Excel archive safetyBlocks workbooks above 20 MB expanded size, 1,024 ZIP entries, 8 MB largest worksheet XML, or 5 MB shared-strings XML
Excel worksheetThe visible grid is limited to 5,000 rows and 200 columns
Inline WordShows the first page before opening the full preview

These are preview limits, not evidence-coverage guarantees. A partial preview does not establish what is absent elsewhere in the file. If a preview is unavailable, use an authorized download or a smaller review copy.

When processing is incomplete

Check the file’s processing state and available extracted content. Do not treat an unsupported or failed extraction as an empty source. Retry an identified failure when appropriate, or supply a readable export if the original format cannot be processed.

Processing depends on the document, extraction route, and configured service limits. See Ingesting documents for duplicate handling and failures.