paper-analyze

Provide an arXiv ID, title, or file path, and the workflow downloads the PDF and TeX source, extracts metadata and figures, then generates a structured Markdown note in your Obsidian vault under 20_Research/Papers with frontmatter, analysis sections, and image references.

It accepts plain arXiv IDs like 2402.12345, prefixed forms like arXiv:2402.12345, paper titles, or direct paths to existing notes. It first searches the vault for existing notes before downloading new content.

Yes. Copy the local PDF into the /tmp/paper_analysis working directory and run the analysis script with the --local-pdf flag, which skips the arXiv download steps and processes the local file directly.

Figures are extracted from the arXiv source package and saved to an images directory under the paper's note folder, with an index.md file. Notes reference figures using relative paths, and PDF-format figures render directly in Obsidian.

The scripts require OBSIDIAN_VAULT_PATH to locate your vault, or you can pass the path via the --vault argument. Without it, the note generation and graph update scripts exit with an error.

The workflow handles errors gracefully: if arXiv is unreachable it uses cached content or notes the limitation, and if the knowledge graph update fails the analysis continues without updating the graph.