Aria is a Zotero plugin designed to enhance academic research by integrating Large Language Models (LLMs) for tasks like referencing, autocompletion, and visual analysis of research papers. It targets researchers and academics using Zotero as their reference manager, aiming to streamline the research workflow by leveraging AI capabilities directly within their existing tool.

How It Works

Aria functions as a Zotero add-on, connecting to OpenAI's GPT-4 model family. It allows users to interact with their Zotero library by dragging and dropping items, authors, or tags into chat prompts. The plugin supports autocompletion for these elements and offers visual analysis capabilities using GPT-4 Vision for PDF content. Chat sessions can be saved as Zotero notes and annotations.

Quick Start & Requirements

• Installation: Download the .xpi file from the latest release and install via Zotero's Add-ons Manager (Tools > Add-ons > Gear Icon > Install Add-on From File). Separate releases are available for Zotero 6 and Zotero 7.
• Prerequisites: Requires an OpenAI API key and access to GPT-4 models. GPT-4 Vision is needed for visual analysis features.
• Setup: Requires adding an OpenAI API key via Zotero preferences and restarting Zotero.
• Documentation: Installation walkthrough available via Twitter link.

Highlighted Details

• Drag-and-drop functionality for referencing Zotero items and collections.
• Autocompletion for authors, tags, and items.
• Visual analysis of PDFs using GPT-4 Vision.
• Chat sessions can be saved as Zotero notes and annotations.

Maintenance & Community

The project is actively maintained, with releases for Zotero 6 and Zotero 7. Users are encouraged to report issues via GitHub issues.

Licensing & Compatibility

The repository does not explicitly state a license in the provided README. Compatibility is limited to Zotero versions 6 and 7.

Limitations & Caveats

Currently, Aria can only query the Zotero library via its search API, not directly access the SQLite database for metrics. Awareness of Zotero's application state (selected item, highlighted text) is limited, requiring manual context provision via drag-and-drop or autocompletion. Some advanced configurations might require manual editing of Zotero's configuration editor.