Skip to main content
Extracts tabular data from PDF/Office/image files into a formatted .xlsx. Display name “PDF to Excel”. Off by default. Never auto-suggested — enable deliberately. This output is already presentation-ready; don’t reformat it with another AI step.

​
Authentication and enablement

No integration. Off by default; opt in per agent.

​
Inputs

  • file_urls (required): 1–5 source URLs per call. If the user gave more than 5, batch them yourself (do not ask). Supported: PDF, PPT/PPTX, DOC/DOCX, XLS/XLSX, CSV, images.
  • guidelines (optional): domain hints (“line-items table”, “each file is one month”).
  • mode — only doc_direct (default). Large PDFs are split into page batches internally.

​
Output

Generated Excel file URL and file id.

​
Limits and side effects

  • Max 5 files per invocation.
  • Writes one spreadsheet to task storage.
  • Free-tier orgs may hit a page cap.

​
Expected errors

  • Empty file_urls, or more than 5 in one call.
  • Unreadable / unsupported file.