| name | pdf |
| description | Read, extract, OCR, create, and edit PDF documents with the installed PDF toolchain. Use when a task directly involves a PDF file or requires a PDF deliverable; do not use for Word documents, spreadsheets, or generic text output. |
PDF documents
Determine whether the PDF is digital text or scanned imagery before choosing an extraction path.
Read
Use pdf-parse for digital PDFs. Inspect a small sample first; if meaningful text is absent, use the available OCR path rather than returning empty output. Preserve page boundaries and cite page numbers when the task depends on location. See reading.md for extraction and OCR patterns.
Create or edit
Use jspdf for straightforward text-first PDFs. For visually complex or source-layout-sensitive work, use the product’s dedicated PDF/rendering surface when available. Keep content inside page bounds, repeat headers intentionally, and use page breaks rather than clipping. See writing.md for generation patterns.
Verify
Confirm the file exists, has a nonzero size, parses successfully, contains expected text, and has the expected page count. Render pages for visual QA when layout matters. Report the exact output path and any OCR/layout limitation.