Claude skill: automated paper-to-article workflow & prompt automation

A Claude skills workflow was built to convert academic papers into polished articles and publish drafts. - The system accepts a paper URL as input. - It generates a Markdown article per specified prompts. - A second pass reflects on reasoning and polishes the text. - The workflow screenshots paper figures for article images. - Nano Banana Pro is used for raw image processing. - It integrates with Obsidian NoteToMP for one-click draft publishing to a public account. - An associated experiment automates a manual slide-by-slide paste step using two faster methods.
Key insights
Claude skills converts paper URLs into Markdown articles
A Claude skills implementation accepts a paper URL and produces a Markdown article according to specified prompts. The workflow includes an initial article generation pass and a second-pass reflection that revises reasoning and polishes the text. The system also automatically extracts and screenshots figures from the paper to use as accompanying images and uses Nano Banana Pro for raw image processing. The final output can be prepared as a public-account draft.
Integration supports one-click publish to public account drafts
The Claude skills workflow integrates with an Obsidian plugin (NoteToMP) to enable one-click publishing of generated articles to a public account draft workspace. The pipeline automates figure capture, image generation, and draft export, streamlining publication from paper URL to platform draft.
Prompt step automation experiments seek faster slide handling
An experiment attempted to automate a manual step in a slide-prompt workflow that originally required pasting each slide's content description sequentially. The author tried two faster methods to replace the manual per-slide paste workflow, reporting promising results and aiming to fully automate the third step of the original prompt sequence.