zotero-read

Zotero Read

Extract and present content from a specific item in the user's Zotero library. This skill is about reading — getting precise content with page/section references.

Prerequisites

• zot CLI: Required for targeted page extraction. Install with uv tool install zotero-cli-cc.
• MCP zotero server: Must be configured and running.

Step 1: Identify the item

The user should have already identified which item (via zotero-find or by naming it). If not clear, ask.

Resolve to an item key:

mcp__zotero__zotero_search_items(query="Hesthaven Warburton", limit=3)

Step 2: Get structure and annotations (in parallel)

Always get these before reading content — they tell you where to look:

# Run all three in parallel:
mcp__zotero__zotero_get_item_metadata(item_key="<KEY>")
mcp__zotero__zotero_get_pdf_outline(item_key="<KEY>")
mcp__zotero__zotero_get_annotations(item_key="<KEY>")

Why annotations first: if the user has highlighted relevant passages, those are the fastest path to the answer — no page scanning needed.

The outline gives PDF page numbers (may differ from printed page numbers). Use PDF page numbers with zot pdf --pages.

Step 3: Extract content — two-phase approach

Short papers (< 20 pages)

Use fulltext — fast and gets everything:

mcp__zotero__zotero_get_item_fulltext(item_key="<KEY>")

Large documents (books, long papers) — Phase 1: targeted extraction

Scan the entire outline for sections relevant to the user's question. A topic may appear in multiple chapters (e.g., boundary conditions in Chapter 6 for hyperbolic AND Chapter 7 for elliptic problems).

Prioritize and limit: Pick the top 3-4 most relevant sections (max ~30 pages total). Extract these first:

zot pdf <KEY> --pages 207-215

If multiple sections are relevant, extract them in parallel using multiple tool calls (not subagents).

Present these results to the user first. If the user wants more depth, they can ask — then extract additional sections. This two-phase approach avoids burning 100k+ tokens on content the user may not need.

Phase 2: deeper extraction (on request)

If the user says "tell me more" or "what about section X", extract additional page ranges. The outline and Phase 1 results inform where to look next.

Step 4: Present with references — separating source from summary

The user must always distinguish between what the document says and your interpretation. Mixing these can present hallucinated content as real.

Structure every finding as:

1. Direct quote — blockquote with exact page reference:

"The exponential filter is applied in the modal space, damping the highest-order modes..." — Hesthaven & Warburton (2008), Section 6.1, p. 134

2. Your summary — clearly labeled:

Summary: The authors use modal filtering to stabilize aliasing-driven instabilities. The filter transfer function (Eq. 6.3, p. 134) controls which modes are damped.

Rules

• Quote first, summarize second. The quote is evidence; the summary is your reading.
• Never fabricate quotes. If you can't find exact wording, say "The authors discuss [topic] in Section X.Y (pp. N-M)".
• If uncertain, say so. "I could not find a direct discussion of X" is better than guessing.
• Page numbers must be verifiable. Only cite pages from content you actually extracted.
• Include equation/figure/table numbers when they exist.

Format:

• (Author Year, Section X.Y, p. N) for single pages
• (Author Year, Section X.Y, pp. N-M) for ranges

Step 5: Citations on request

mcp__zotero__zotero_get_item_metadata(item_key="<KEY>", format="bibtex")

Or: zot export <KEY> / zot export <KEY> --format csl-json

Tools reference

Tool	Purpose
zotero_search_items	Resolve title/author to item key
zotero_search_by_citation_key	Resolve citekey to item key
zotero_get_item_metadata	Metadata, abstract, BibTeX
zotero_get_pdf_outline	Table of contents with page numbers
zotero_get_item_fulltext	Full text (short papers only)
zotero_get_annotations	User's highlights and notes
zot pdf <KEY> --pages X-Y	Extract specific pages (large docs)
zot export <KEY>	Export BibTeX/CSL-JSON/RIS

Tips

• zot pdf caches extracted text locally — repeated reads are instant.
• Annotations are the fastest signal — if the user highlighted something, start there.
• For broad topics in large books, extract the top 3-4 sections first (~30 pages max), present results, and offer to go deeper. Don't extract everything upfront.
• If the outline is missing, try the first few pages for a table of contents, or use annotations as navigation.