yilewang/llm-for-zotero

What it solves

llm-for-zotero integrates Large Language Models (LLMs) directly into the Zotero PDF reader, eliminating the need to switch between your research library and an AI assistant. It solves the friction of manually summarizing papers, extracting evidence, and organizing research notes across different applications.

How it works

The plugin acts as a research assistant within Zotero, connecting to various LLM backends (API keys, local models, WebChat, Codex App Server, or Claude Code). It provides grounded chat capabilities by loading the current PDF as context and using retrieval-augmented generation (RAG) to provide answers with citations that link back to the source passage.

For advanced workflows, it features an Agent Mode that can interact with the Zotero library via read/write tools to search items, update metadata, apply tags, and import new papers. It also integrates with Tavily for general web search and MinerU for high-fidelity PDF parsing of tables and equations.

Who it’s for

Researchers, academics, and students who use Zotero to manage their literature and want to leverage AI to accelerate their reading, synthesis, and organization of scholarly papers.

Highlights

• Grounded Paper Chat: Chat with PDFs, figures, and selected text with one-click jumps to source citations.
• Agent Mode: A beta feature allowing the AI to search the library, manage metadata, and automate organization workflows.
• Multi-Paper Comparison: Compare multiple open papers using slash commands to add context.
• External Note Integration: Save AI-generated research notes directly to Zotero or local Markdown folders like Obsidian and Logseq.
• Flexible Backends: Supports standard API providers, local OpenAI-compatible models, and specialized bridges like Codex App Server and Claude Code.
• Citations & Discovery: Integrated scholarly discovery via CrossRef and Semantic Scholar, plus general web search via Tavily.