github 54yyyu/zotero-mcp v0.3.0

8 hours ago

What's New

New Tools

  • Note CRUDzotero_update_note and zotero_delete_note for full note lifecycle management, plus raw_html parameter on note readers for round-trip editing (#199)
  • PDF area annotationszotero_create_area_annotation for creating image/area annotations on PDFs using normalized coordinates (#190)

New Features

  • Gemini embedding v2 — support for gemini-embedding-2-preview (3072-dim, 8192-token context) with in-prompt task prefixes and batched embedding (up to 100 per call) (#205)
  • 10 new update_item fieldsvolume, issue, pages, publisher, issn, language, short_title, edition, isbn, book_title, with skip warnings for inapplicable fields (#195)
  • Related items in markdown## Related Items section now included in format_item_metadata output (#191)
  • Collection-scoped searchcollection_key parameter on search/retrieval tools to scope queries to specific collections (#194)
  • Community installer — linked ehawkin/zotero-mcp-setup in README for users who prefer a GUI installer

Bug Fixes

  • Annotation hierarchy traversalzotero_get_annotations now correctly descends through PDF attachments with content-type filtering and deduplication (#201, #194)
  • Annotation key resolution — proper routing for attachment vs. parent item keys, with parent title fallback resolution (#194)
  • tiktoken special tokens — fixed ValueError on <|endoftext|> during fulltext indexing (#188)
  • Group library 404library_type now correctly pluralized for pyzotero write client (#200)
  • ctx.warn → ctx.warning — fixed AttributeError in error-handling branches where ctx.warn() was called but fastmcp.Context only defines warning() (#202)
  • Config merge precedence — env vars no longer silently overwrite config.json values (#204)
  • Query truncationembed_query now truncates to max_input_tokens before API call (#204)
  • Semantic stats accuracy — recovered docs no longer inflate added_items count (#204)
  • API key propagationbuild_from_config now passes api_key for OpenAI and Gemini (#204)
  • chromadb ImportError — clear, actionable error message when chromadb is missing (#189)
  • Hazardous docs link — fixed README/pyproject.toml links from expired .us domain to .com (#197, #206)
  • pyzotero version — bumped minimum to >=1.6.0 to ensure local kwarg support (#203)
  • Test reliability — fixed mock patching for Python 3.11/3.12 compatibility in CI

Full Changelog: v0.2.2...v0.3.0

Don't miss a new zotero-mcp release

NewReleases is sending notifications on new releases.