Summary

This project provides a "Codex skill" designed to automate the creation of image-based PowerPoint presentations from various text inputs like articles, reports, or notes. It targets users of agent frameworks such as Codex, Claude Code, OpenClaw, and Hermes Agent, offering a streamlined workflow for generating visually rich slides that are assembled into a final .pptx file. The primary benefit is the automated generation of aesthetically consistent presentations, ideal for strong visual communication needs.

How It Works

The skill operates by first parsing input content to plan a presentation outline and define a visual style. It then leverages AI image generation models, prioritizing Codex's built-in capabilities or falling back to gpt-image-2 via local API/CLI configurations. Each slide is rendered as a complete 16:9 image. Finally, a local Python script assembles these generated images into a standard .pptx file, ensuring a unified visual language across the presentation. This approach allows for flexibility in image generation backends and supports user-defined visual styles.

Quick Start & Requirements

• Installation: Use npx -y skills@latest add ningzimu/codex-ppt-skill for Codex, Claude Code, and Hermes Agent, or openclaw skills install codex-ppt for OpenClaw. Local development can utilize symbolic links.
• Prerequisites: Compatibility with SKILL.md-based agents. For non-Codex environments or when using fallback image generation, configuration of OPENAI_API_KEY (and optionally base-url, model) in ~/.codex-ppt-skill/.env is required.
• Links: ClawHub for OpenClaw. A QR code links to a Skill Exchange Group for community support and updates.

Highlighted Details

• Supports multiple agent environments (Codex, Claude Code, OpenClaw, Hermes) with flexible image generation backends.
• Generates full-page, 16:9 image-based slides for strong visual impact.
• Includes a library of pre-defined visual styles (e.g., "Clean Professional", "Research Defense", "E-ink Magazine") and allows custom style integration.
• Can incorporate user-specified image assets (figures, diagrams) into designated slides.
• Automates the assembly of generated slide images into a .pptx file using a local script.

Maintenance & Community

The project encourages user customization and contribution of preferred visual styles to the references/ directory. A community "Skill Exchange Group" is available via QR code for user support, feedback, and update notifications.

Licensing & Compatibility

The project is released under the MIT License, permitting broad use, modification, and distribution, including for commercial purposes. It is designed for compatibility with agents supporting the SKILL.md format.

Limitations & Caveats

The generated slides are image-based and not directly editable within the PowerPoint file; a separate skill is recommended for conversion to editable formats. The skill's generalized design for multi-agent compatibility may introduce complexity or potential instability for specific workflows, with users encouraged to tailor it. API key configuration is mandatory for fallback image generation, which may present an initial setup hurdle.