Skip to content

KNOWN_ISSUES.md

Latest commit

 

History

History
58 lines (39 loc) · 2.26 KB

KNOWN_ISSUES.md

File metadata and controls

58 lines (39 loc) · 2.26 KB
Raw

Known Issues

No OCR Support for Images

Despite some documentation suggesting OCR support, the MarkItDown library does not perform OCR on images. When converting image files (PNG, JPG, etc.), it only:

  • Extracts EXIF metadata (if available)
  • Can describe images using a multimodal LLM (if configured with an LLM client)

Current behavior: Screenshots and images with text will convert to empty or metadata-only Markdown files.

Workaround: If you need OCR functionality, you would need to:

  1. Use a separate OCR tool (like tesseract) to extract text first
  2. Or wait for OCR support to be added to the upstream MarkItDown library

Unicode Errors with Markdown Files

Some .md files may fail to convert with a UnicodeDecodeError when they contain non-ASCII characters. This is a known limitation in the MarkItDown library's PlainTextConverter, which uses ASCII encoding by default.

Workaround: Convert these files using a different method or wait for an upstream fix in the MarkItDown library.

Affected error:

PlainTextConverter threw UnicodeDecodeError with message: 'ascii' codec can't decode byte...

Office File Dependencies

Excel (.xlsx, .xls) and PowerPoint (.pptx) files require additional dependencies. If you see errors like:

XlsxConverter threw MissingDependencyException...
PptxConverter threw BadZipFile...

Solution: Install the required dependencies using pipx:

pipx inject markitdown-mcp openpyxl xlrd pandas tabulate

Corrupted Office Files

Some Office files may fail with BadZipFile errors if they are corrupted or use an incompatible format. This typically happens with files that:

  • Are password protected
  • Use very old Office formats
  • Are corrupted or partially downloaded
  • Were created with non-Microsoft Office software

Solution: Try opening and re-saving the file in Microsoft Office or a compatible application.

Large File Timeouts

Very large files (especially PDFs with many pages) may cause timeouts in Claude Desktop. The default timeout is 2 minutes per file.

Solution: For batch conversions of large files, consider:

  • Converting smaller batches at a time
  • Using the tool directly from command line for very large files
  • Splitting large PDFs into smaller chunks before conversion