Present Next

Stories, designed to move.

Present Next is an open-source Codex plugin for designing, presenting, and precisely editing expressive HTML presentations. It keeps Zara Zhang's excellent show-don't-tell workflow—three real visual directions first, one coherent design system after selection—and extends it with enhanced design assets, more native motion mechanisms, bilingual typography, a PowerPoint-style presenter workbench, and reversible micro-edits.

Current release: 0.2.0 Preview · one plugin · three focused Skills · English-first with Chinese and mixed-language support

Quick Start

codex plugin marketplace add xuan-labs/present-next
codex plugin add present-next@present-next

Restart required: After installation or an upgrade, fully quit and reopen Codex, then start a new task. Codex indexes Skills when a session starts, so $present-next-preview, $present-next-presenter, and $present-next-edit may not appear in autocomplete inside a task that was already open.

After restarting Codex, start a new task and try:

$present-next-preview

Create a 12-slide launch presentation for an AI research product.
Use English, a speaker-led structure, and balanced motion.

Present Next will open three real title-slide directions, wait for your choice, then generate and validate the complete fixed-stage HTML deck.

If the Skills still do not appear after restarting, run codex plugin list to confirm that present-next is installed, then open another new task.

Three Skills, One Install

Skill	What it adds	When it loads
$present-next-preview	Visual discovery, richer design assets, native presentation mechanisms, motion, bilingual HTML deck generation	When creating or rebuilding a deck
$present-next-presenter	Current/next view, persistent and AI-ready notes, timers, rehearsal, display controls, shortcuts, camera and local recording	Only when Presenter is requested
$present-next-edit	One-element text, typography, color or effect changes with verification and undo	Only when a narrow edit is requested

The three Skills share one deck protocol, stable slide IDs, element IDs, notes, and extension slots. Their implementations remain separate so ordinary deck generation does not pay the context cost of Presenter or Edit.

What It Does

Present Next helps users create a web presentation without choosing abstract design settings up front. It follows a show-don't-tell flow:

1. Ask the missing content questions together.
2. Generate and open three real title-slide directions.
3. Let the user choose Style A, B, C, or a specific mix.
4. Load exactly one full design system.
5. Generate, validate, and open the complete fixed-stage HTML deck.
6. Offer Presenter, narrow Edit, PDF export, or publishing as follow-up actions.

The current package includes:

• 19 validated Zara Enhanced design systems;
• 24 accepted native presentation mechanisms;
• 19 design-system mechanism adapters;
• richer organization, data, relationship, state, and motion expressions;
• English-first typography with Chinese and mixed-language support;
• fixed 1920×1080, 16:9 geometry;
• motion and Reduced Motion behavior;
• an optional PowerPoint-style dual-window Presenter;
• persistent notes with an interface reserved for AI-generated notes;
• precise, reversible one-element edits with screenshot and geometry checks;
• deterministic build, layout, contrast, state, and mechanism checks.

Usage

Invoke the Skill explicitly:

$present-next-preview

Help me create a presentation about how an AI product team moves from
impressive demos to repeatable delivery. I only have the topic.

The Skill will ask the missing purpose, length, density, language, and motion questions together. It then creates and opens three real visual directions and waits for a choice before producing the complete deck.

The user should not need to know about Slide Jobs, mechanism routing, context traces, or build protocols. Those run behind the experience.

Input Modes

• New presentation — Build from a topic, rough notes, Markdown, or complete content.
• PowerPoint conversion — Extract and confirm text, images, order, and notes before visual discovery.
• Existing HTML enhancement — Improve an existing fixed-stage deck while protecting content and geometry.

Visual Discovery

The Skill reads a compact index, selects three materially different candidates, and loads only their lightweight preview.md cards.

After the user chooses, it loads exactly one design.md. The chosen design system governs the entire deck—not just the cover.

Each fixed system follows Zara's two-file asset contract:

templates/<slug>/
├── preview.md
└── design.md

Native Mechanisms

Mechanisms are optional page-expression assets for system maps, metrics, comparisons, decisions, and other narrative jobs.

They preserve organization, relationships, state, motion paths, timing, and terminal meaning while being materialized through the selected design system. The final slide never mounts an application UI component.

Each deck uses at most three mechanisms, and only when the content benefits.

Progressive Loading

Normal generation loads:

1. one compact selection index;
2. at most three Preview Cards;
3. exactly one full Design System;
4. only the few Mechanism Specs and adapters selected for relevant pages.

It does not load all design systems, all component prompts, or the complete mechanism inventory into model context.

Repository Structure

present-next/
├── .agents/plugins/marketplace.json
├── plugins/present-next/
│   ├── .codex-plugin/plugin.json
│   └── skills/
│       ├── present-next-preview/
│       ├── present-next-presenter/
│       └── present-next-edit/
├── README.md
└── THIRD_PARTY_NOTICES.md

Upstream

The governing workflow and part of the fixed visual-system foundation derive from Zara Zhang's frontend-slides, pinned during development to v2.1.0 at commit 9906a34d640d2111f724544cbc50f7f130569ae1.

Present Next keeps Zara's visual-discovery, fixed-stage, Preview/Design, deck generation, delivery, and export architecture. It extends the asset dimension and validation layer rather than replacing that workflow.

See:

• plugins/present-next/skills/present-next-preview/assets/zara/LICENSE
• THIRD_PARTY_NOTICES.md
• plugins/present-next/skills/present-next-preview/references/upstream-and-rights.md

Release Status

Version 0.2.0 is a bundled Preview release:

• zara-enhanced is the only exposed generation mode;
• original-fixed and generative-compose are not released;
• 19 systems are active;
• Presenter and Edit are bundled but loaded only on explicit requests;
• the package remains evaluation-only until all bundled mechanism sources have explicit redistribution metadata.

License

Original Present Next code and documentation are available under the MIT License. Third-party and derived materials retain their own notices and may have additional restrictions. See LICENSE and THIRD_PARTY_NOTICES.md.