llm-for-zotero is a Zotero plugin that integrates large language models into the Zotero PDF reader to enable research workflows.
The tool solves the friction of switching between a reference library and AI tools by embedding LLM capabilities directly into Zotero. It lets researchers ask questions about papers, summarize content, analyze figures, compare sources, and save notes without leaving their library. The system works with multiple backend options including standard API providers, local OpenAI-compatible models, WebChat, Codex App-Server, and Claude Code. Answers include citations that link back to source passages in the PDF.
Researchers managing large paper collections in Zotero should consider this tool if they want to reduce context-switching during literature review. The plugin supports comparing multiple open papers simultaneously and adding external files as context. It can save conversations and research notes either to Zotero's native note system or to external Markdown folders like Obsidian and Logseq. An Agent Mode enables library-wide operations including search, tagging, metadata editing, and import workflows. The tool also integrates web search through Tavily to ground answers in current public information. The flexibility of backend selection means researchers can use their preferred LLM provider or run local models for privacy.
The project shows active development with documentation available in multiple languages and a roadmap section indicating planned features. The codebase is written in TypeScript and includes experimental features like Claude Code integration alongside stable functionality. The README documents multiple setup paths for different user preferences, from API-based approaches to self-hosted solutions, suggesting the maintainers are responsive to diverse deployment scenarios.
