917Dhj/DeepPaperNote

What it solves

DeepPaperNote addresses the friction of turning complex academic papers into durable, usable notes. It eliminates the repetitive manual work of switching between PDFs, reference managers like Zotero, and note-taking apps, preventing the common problem where reading a paper results in fragmented notes or disposable AI summaries that lack theesprit de corps of the original research.

How it works

It functions as an agent skill (compatible with Claude Code and Codex) that processes one paper at a time. The system extracts research questions, methods, evidence, results, and figures from a PDF or reference (via DOI, URL, or arXiv ID) and structures them into a formatted Obsidian note. It uses PyMuPDF for PDF extraction and can integrate with Zotero's local API and Semantic Scholar for better metadata and file retrieval.

Who it’s for

Researchers, students, and academics who use Obsidian as a personal knowledge base (academic wiki) and want to automate the ingestion of research papers into their system without losing the technical depth (formulas, numbers, and figure context) of the original work.

Highlights

• Obsidian Integration: Specifically designed to produce notes that are searchable, linkable, and reusable within an Obsidian vault.
• Evidence-Based: Focuses on preserving formulas, experimental conclusions, and figure context rather than providing a generic summary.
• Flexible Input: Supports local PDFs, DOIs, URLs, arXiv IDs, and Zotero items.
• Companion Skills: Includes an optional paper-glossary skill to create reusable terminology notes from paper artifacts.
• Multi-Agent Support: Distributed as a plugin for various AI agents, allowing the to be integrated into existing LLM workflows.