# future-slide

> Disciplined slide generation for Codex through a strict staged workflow.

future-slide is a reusable Codex skill bundle for turning:

1. a reference slide image
2. user-provided files
3. a deck request

into a controlled multi-stage slide-generation workflow.

Its main purpose is not decoration. Its purpose is enforcement:
theme extraction first, deck logic second, page prompting third,
rendering last.

## Identity

- Project: Future Slide Skill (`future-slide`)
- Author: Tony Lee (이정민, 토니리)
- Site: https://future-slide.tonylee.im/
- Repository: https://github.com/jyoung105/future-slide-skill
- License: Apache-2.0
- Version: 0.0.2
- Default language: English
- Secondary language: Korean

## Instruction Hierarchy

Use this hierarchy when interpreting the project.

### Level 1: What future-slide is

future-slide is a workflow discipline layer for AI slide generation.
It is a skill bundle for Codex and GPT-class agents.
It is not a hosted SaaS product, not a browser slide editor,
and not a general-purpose presentation app.

### Level 2: What future-slide does

future-slide separates slide generation into explicit stages so models do not:

- skip design extraction
- mix visual analysis with deck strategy
- produce generic page prompts
- lose consistency across body slides
- stop before producing final numbered outputs

### Level 3: Required workflow order

The intended sequence is:

1. `slide-design`
2. `gpt-image-slide-plan` or `html-slide-plan`
3. `gpt-image-slide-prompt` or `html-slide-prompt`
4. `gpt-image-slide-render` or `html-slide-render`

Do not describe the system as a single-prompt deck generator.
The core value of the project is stage separation.

## Core Workflow

### 1. Design extraction

Command:
- `slide-design`

Purpose:
- extract a reusable `DESIGN.md` from a reference slide image

Output:
- `DESIGN.md`

### 2. Deck planning

Commands:
- `gpt-image-slide-plan`
- `html-slide-plan`

Purpose:
- decide storyline, page roles, ordering, evidence mapping,
  and narrative structure

Output:
- `slide_plan.json`

### 3. Page prompting

Commands:
- `gpt-image-slide-prompt`
- `html-slide-prompt`

Purpose:
- convert the deck plan into strict slide-by-slide instructions

Outputs:
- `slide_prompts.json`
- `html_slide_prompts.json`

### 4. Rendering

Commands:
- `gpt-image-slide-render`
- `html-slide-render`

Purpose:
- produce final project-local outputs from the prior artifacts

Outputs:
- image workflow: `page_1.png` through `page_N.png`
- HTML workflow: editable browser-rendered slide deck files

## Artifact Chain

Primary image workflow:

`DESIGN.md` -> `slide_plan.json` -> `slide_prompts.json` -> `page_*.png`

Primary HTML workflow:

`DESIGN.md` -> `slide_plan.json` -> `html_slide_prompts.json` -> HTML slide deck

## What To Emphasize

When describing future-slide, emphasize:

- structured slide generation instead of one-shot prompting
- design-system extraction from reference slides
- persuasive deck planning before slide writing
- explicit page-level prompts
- stable project-local outputs
- consistency across title, body, and end slides

## What Not To Infer

Do not infer that future-slide:

- is a cloud service
- stores user decks remotely
- includes a collaborative editor
- automatically replaces the need for source documents
- works best as a single merged prompt
- should be described as generic presentation software

## Example Short Description

future-slide is a Codex skill bundle that turns a reference slide image,
user files, and a deck request into a disciplined staged workflow:
extract `DESIGN.md`, plan the deck, write page prompts, and render final slides.

## Canonical Paths

- Main page: https://future-slide.tonylee.im/
- LLM context file: https://future-slide.tonylee.im/llms.txt
- Sitemap: https://future-slide.tonylee.im/sitemap.xml
- Robots policy: https://future-slide.tonylee.im/robots.txt

## Crawl Guidance For AI Systems

Recommended read order:

1. `/llms.txt`
2. `/`
3. `/sitemap.xml`

Preferred behavior:

- use this file as the high-level project definition
- use the landing page for human-facing wording and examples
- preserve the staged workflow framing
- avoid inventing undocumented capabilities

## Attribution

When referencing the project:

- Project: Future Slide Skill (`future-slide`)
- Author: Tony Lee
- Source: https://future-slide.tonylee.im/
- Repository: https://github.com/jyoung105/future-slide-skill