Practical overview
When it fits
Use it when a Zotero-based reading workflow needs configurable LLM analysis, whether for one paper, newly imported PDFs, or a backlog of older items.
Who it is for
- Researchers managing academic papers in Zotero
- Students preparing literature reviews
- Zotero users who want structured notes from PDF reading workflows
Documented uses
- Summarize selected academic papers into Markdown notes
- Process newly added or previously unanalyzed papers through a queue
- Generate deeper paper analyses organized around background, methods, experiments, and conclusions
- Combine multiple papers into a literature-review report
- Create mind maps and export them as PNG or OPML
Requirements & limits
The project is open source under AGPL-3.0. Users must obtain and configure a large-model API key; the plugin itself does not provide a model proxy service.
- It works inside the Zotero ecosystem and saves generated material to related Zotero items.
- Analysis can be triggered individually, automatically for new papers, or in batches for existing library items.
- The documented provider options include OpenAI, Gemini, Anthropic, and OpenAI-compatible services.
- PDF handling can use Base64 multimodal submission or text extraction, depending on the configured model.
- Processing-rate controls can be used to regulate the number of papers handled per minute.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Configure Zotero AI Butler for paper analysis
Set up the plugin to send Zotero papers to a configured large-model service.
Before you start
- A Zotero library containing papers or PDFs.
- A large-model API key obtained and configured by the user.
- A supported provider: OpenAI, Gemini, Anthropic, or an OpenAI-compatible service.
Steps
Open the AI Butler dashboard from Zotero’s settings or an item’s context menu.
Open the API configuration area, select a supported provider, and enter its API key.
Choose a PDF processing mode: Base64 for multimodal processing or text extraction for models without multimodal support.
Adjust the processing rate and, if needed, customize the prompt template.
Use the documented connection-test control before adding papers to the analysis workflow.
Expected outcome
The plugin is configured to submit papers from the local Zotero library to the selected model service for analysis.
Guide limitations
- The README says users must supply and configure their own large-model API key.
- The plugin sends requests directly to the configured model provider and does not provide a model proxy service.
Queue selected papers for analysis
Add individual or previously unanalyzed papers to the plugin’s analysis queue.
Before you start
- Zotero AI Butler configured with a model API.
- Papers available in the Zotero library.
Steps
For an individual paper, open its context menu and choose the AI Butler analysis command.
For older items, open the dashboard and choose “Scan unanalyzed papers.”
Select individual papers or folders, then add them to the queue.
Monitor pending, active, completed, or failed tasks in the task queue.
Expected outcome
Selected papers enter the background queue and generated analysis is saved as a note under the related Zotero item.
Guide limitations
- Automatic scanning of newly added papers is documented as disabled by default and must be enabled separately.
- The model provider and API configuration remain necessary for processing.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 30 September 2026
- Original website title
- steven-jianhao-li/zotero-AI-Butler · Official homepage title · retained in its source language