Upgrade Storage[Plugin] AI Summary for Zotero: Save AI-generated paper summaries
https://s3.amazonaws.com/zotero.org/images/forums/u12291920/gqtetl4g4hutzlfe3fwj.png
Hi everyone,
I recently built a new tool called AI Summary for Zotero, and I wanted to share it with the community.
If you use AI CLIs (like Claude Code, Codex, or Gemini) to help read and analyze papers, you might find yourself dealing with loose Markdown files. I wanted a streamlined way to turn that AI output directly into durable, structured Zotero child notes attached right beside the paper in my library.
How it works:
It combines a lightweight local Zotero plugin with two AI CLI skills (/summary and /pinpoint-lesson). The agent extracts evidence from your PDF, drafts the note, and sends it to the plugin via a local HTTP API, which then saves it as a child note under the matched Zotero item.
Key Features:
Zotero-Native Notes: Summaries and deep lessons are saved directly as child notes under existing Zotero items.
Figure Extraction: Automatically extracts raster figures from the PDF (using Poppler) and embeds them in the note.
Math Rendering: Converts inline LaTeX formatting to KaTeX so equations render perfectly in Zotero's built-in note viewer.
Cross-CLI Support: The skills work across Claude Code, OpenAI Codex CLI, and Google Gemini CLI.
Requirements: Zotero 9.0+, Poppler, and Python 3.
You can find the installation guide, source code, and examples here:
https://github.com/ajs7270/zotero-ai-summary
I built this to scratch my own itch, but I'm hoping it's useful to other researchers here. I'd love to hear your feedback, feature requests, or any issues you run into!> Hi everyone,
Hi everyone,
I recently built a new tool called AI Summary for Zotero, and I wanted to share it with the community.
If you use AI CLIs (like Claude Code, Codex, or Gemini) to help read and analyze papers, you might find yourself dealing with loose Markdown files. I wanted a streamlined way to turn that AI output directly into durable, structured Zotero child notes attached right beside the paper in my library.
How it works:
It combines a lightweight local Zotero plugin with two AI CLI skills (/summary and /pinpoint-lesson). The agent extracts evidence from your PDF, drafts the note, and sends it to the plugin via a local HTTP API, which then saves it as a child note under the matched Zotero item.
Key Features:
Zotero-Native Notes: Summaries and deep lessons are saved directly as child notes under existing Zotero items.
Figure Extraction: Automatically extracts raster figures from the PDF (using Poppler) and embeds them in the note.
Math Rendering: Converts inline LaTeX formatting to KaTeX so equations render perfectly in Zotero's built-in note viewer.
Cross-CLI Support: The skills work across Claude Code, OpenAI Codex CLI, and Google Gemini CLI.
Requirements: Zotero 9.0+, Poppler, and Python 3.
You can find the installation guide, source code, and examples here:
https://github.com/ajs7270/zotero-ai-summary
I built this to scratch my own itch, but I'm hoping it's useful to other researchers here. I'd love to hear your feedback, feature requests, or any issues you run into!> Hi everyone,