Summarize a PDF into a note

The problem

You have a 60-page report attached to a Zotero item and you want a structured summary you can find later — living on the item, not in a separate app. Zotero has no built-in “summarize this attachment,” and copying text out to an LLM by hand loses the link back to the reference.

Via an AI agent

“Summarize the PDF on item @ojedaDigitalizacion2025 chapter by chapter, keep page references, fold in my existing highlights, and save it as a note on the item.”

The agent resolves the item, gets the PDF path, reads it directly (in page ranges), summarizes bottom-up so each step fits in context, incorporates your annotations, and writes the result back as a child note.

Via the CLI

The CLI locates the PDF and manages the note; you (or your agent) do the reading and writing of prose.

zot pdf @ojedaDigitalizacion2025        # -> local path to the PDF
zot annotations @ojedaDigitalizacion2025 # your highlights (what you flagged matters)

Read the PDF file directly with a PDF-capable tool — only the page ranges you need. For long documents, summarize bottom-up: section → chapter → whole, keeping page references at each level. Then save the summary as a child note:

zot note @ojedaDigitalizacion2025 --file summary.html --if-not-exists
zot notes @ojedaDigitalizacion2025       # list the item's notes to confirm

zot annotations <key> --to-note can also turn your existing highlights straight into a note.

Safety net

Notes are low-risk: they are added as child items and are reversible via trash, not permanent erase. Prefer notes over writing PDF annotations back — placing highlights requires fragile coordinate math. Use --if-not-exists to avoid creating a duplicate note if you re-run the workflow. To remove a note, send it to the trash rather than erasing it.