Analysis updated 2026-06-24
Convert a chemistry paper into a Chinese 16:9 PPTX deck for a group meeting using cropped real figures only.
Produce a journal-club deck where each page heading is a conclusion sentence backed by a cited figure.
Prepare a thesis defense rehearsal deck with adaptive section ordering for a materials or survey paper.
Build a figure source manifest that maps every slide image back to the paper or supplementary information.
| fangyuanopus/literature-report-ppt-builder | jun7799/scribe-transcribe | rbrown101010/codex-marketing-skills | |
|---|---|---|---|
| Stars | 40 | 40 | 40 |
| Language | — | Python | Python |
| Setup difficulty | easy | moderate | moderate |
| Complexity | 2/5 | 3/5 | 2/5 |
| Audience | researcher | general | writer |
Figures from each repo's GitHub metadata at analysis time.
Drop the skill folder into ~/.codex/skills or ~/.claude/skills, then run inside Codex or Claude Code with the paper available.
This is a Codex and Claude Code skill, written for a Chinese-speaking audience, that turns an academic paper into a Chinese-language presentation deck. The author calls it academic-slide-minimalist. It is meant for situations where a student walks an advisor or seminar group through a paper page by page, such as a group meeting, a journal club, a course report, or a thesis defense rehearsal. The stated principle is that clarity comes before looks and real evidence before flashy visuals. The skill names the problem it wants to fix. Many AI deck tools produce attractive slides but stack figures in the paper's original order without following the argument, sometimes draw lookalike images that are not real experimental data, use vague topic headings instead of conclusion sentences, and skip Q&A preparation. This skill reverses the priority: lock down the paper's logic, figure sources, and evidence strength first, and only then worry about how the page looks. The workflow runs in a fixed sequence: close reading, a logic tree, a terminology table, a crosswalk between the main paper and supplementary information, a figure source manifest, an adaptive navigation, a deck order map, per-page briefs, image-based page generation, PPTX assembly, and a final quality and Q&A audit. Real figures must come from the paper, the supplementary information, or user-uploaded screenshots. Generating or redrawing experimental images such as crystal structures, spectra, microscopy, or performance plots is forbidden, only cropping, framing, arrows, and short labels are allowed. Pages are built image-first: each 16:9 slide is rendered as a single image and placed into the PPTX, which avoids font swaps when the deck is opened on another machine. Adaptive navigation means the section order changes by paper type, so a materials paper and a survey paper do not get the same outline. Installation has three paths: hand the GitHub URL to the agent and let it install, or clone and copy the academic-slide-minimalist folder into ~/.codex/skills or ~/.claude/skills. The README lists about fifteen intermediate markdown files the skill writes during a run, plus the final PPTX, and warns that the sample deck inside the repo is only a style reference. The license is MIT.
A Codex and Claude Code skill that turns an academic paper into a Chinese-language PPTX deck, locking down logic and real figure sources before any visual styling.
MIT lets you use, modify, and ship this for any purpose, commercial or not, as long as you keep the copyright notice.
Setup difficulty is rated easy, with roughly 30min to a first successful run.
Mainly researcher.
This repo across BitVibe Labs
Verify against the repo before relying on details.