What is future-slide?

future-slide is a reusable Codex skill bundle that turns a reference slide image, user files, and a deck request into a disciplined four-stage workflow: extract DESIGN.md, plan the deck, write per-page prompts, and render slide images sequentially.

How do I install future-slide?

Run `npx skills add jyoung105/future-slide-skill` (Node.js 18+), or copy the four skill folders into ~/.codex/skills/ and restart Codex.

future-slide

v0.0.2 Apache-2.0

Disciplined slide generation, one stage at a time.
Future Slide Skill turns a reference slide image, your files, and a deck request into a strict four-stage workflow, so GPT-class models stop skipping steps and collapsing into generic decks.

View skill

Four-stage future-slide workflow diagram

Four skills, one disciplined pipeline.

01

slide-design

DESIGN.md

Extract a reusable DESIGN.md from a reference slide image: placement, header/body/footer flow, title/body/end-page behavior, icon usage, infographic cards, and diagram rules.

Inputs · Reference slide image(s)

02

gpt-image-slide-plan

slide_plan.json

Build the storyline and page-family rhythm. Decide where tables, charts, icon-led modules, and comparison exhibits belong. Map evidence to slides.

Inputs · DESIGN.md · user files · goal & audience

03

gpt-image-slide-prompt

slide_prompts.json

Convert the plan into strict per-page prompt JSON. Make layout family, zoning, and anti-generic constraints explicit for every page.

Inputs · DESIGN.md · slide_plan.json

04

gpt-image-slide-render

page_1.png ... page_N.png

Render slide images one by one with Codex native image generation. Save project-bound assets explicitly with stable, page-numbered filenames.

Inputs · DESIGN.md · slide_prompts.json

A single prompt fails in predictable ways.

So this bundle separates responsibilities. Theme first. Story second. Page-level prompts third. Generation last, with explicit, numbered outputs.

Starts writing slides before the theme is extracted.
Mixes design analysis with deck strategy.
Produces page prompts without a real narrative arc.
Loses layout consistency across body slides.
Overfits to visible text, not the slide design system.
Stops after writing prompts, never renders numbered slides.

A typical end-to-end run.

Drop a reference image, point at a PDF, and let each skill stay in its lane.

end-to-end

# 1) Extract design system from reference slide
$ slide-design [reference slide image]

# 2) Build the deck plan from a real source document
$ gpt-image-slide-plan /path/to/report.pdf
  Create a full Korean research-summary deck.

# 3) Convert the plan into strict per-page prompts
$ gpt-image-slide-prompt

# 4) Render slides sequentially
$ gpt-image-slide-render

DESIGN.md
→
slide_plan.json
→
slide_prompts.json
→
page_1.png ... page_N.png

Install once. Restart Codex. Ship decks.

With the Skills CLI · Requires Node.js 18+.

terminal

$ npx skills add jyoung105/future-slide-skill

Manual install · Copy each skill folder into your Codex skills directory.

terminal

$ mkdir -p ~/.codex/skills
$ cp -R skills/slide-design ~/.codex/skills/
$ cp -R skills/gpt-image-slide-plan ~/.codex/skills/
$ cp -R skills/gpt-image-slide-prompt ~/.codex/skills/
$ cp -R skills/gpt-image-slide-render ~/.codex/skills/

Stricter enforcement, not more decoration.

Theme extraction first, content second.
Strong cross-slide consistency.
Persuasive story flow with explicit evidence mapping.
Reusable layout families and body-slide discipline.
Header/body/footer zoning with role-clear icons & charts.
Sequential image generation with stable numbering.
No hallucinated rules when the reference is ambiguous.