Skip to content
Get packs

Guide · Agent skill

How to use the Slide Deck Generator agent skill (baoyu-slide-deck)

Updated · written with AI from the project's own documentation and checked against our catalogue data

The Slide Deck Generator is an agent skill, a folder of instructions and scripts that an AI agent such as Claude loads when a task matches its description. This guide explains what the baoyu-slide-deck skill makes the agent do, how to install it, how to ask for a deck in plain language, and what it needs to run. By the end you will be able to turn notes or an article into a planned, shareable set of slide images.

Slide Deck GeneratorJimLiu/baoyu-skills · MIT

Quick start

  1. 1Download and unzip the skill. Download the ZIP from this page and unzip it so the baoyu-slide-deck folder contains SKILL.md.
  2. 2Place the folder. Move the folder into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder for one project. Alternatively upload it under Settings → Capabilities → Skills in the Claude apps.
  3. 3Confirm an image backend. Make sure you have Codex imagegen, Codex CLI with an active login, Cursor GenerateImage, or baoyu-image-gen available.
  4. 4Ask for a deck. Give the agent your content and use a trigger such as "make a presentation" or "create slides".
  5. 5Confirm the plan. Review the outline and style the agent proposes, and reply with changes or approval.
  6. 6Review outputs. Check the generated slide images and the prompt files in prompts/, and ask for regeneration of any slide with wrong text.

What the skill makes the agent do

When installed, the agent takes your notes, article or topic and plans a multi-slide deck. It first writes an outline with style and layout instructions, then generates one image per slide.

The skill is built for slides meant to be read or shared as a scroll, such as self-explanatory decks or social-media carousels, rather than presented live. The output is generated slide images, not editable PowerPoint slides.

  • Input: notes, an article, a file path or a topic
  • Planning: an outline with style and layout instructions
  • Output: one image per slide, plus a saved prompt file for each slide
  • Optional merge into PPTX or PDF using the bundled scripts once all slides exist

Install the skill from the ZIP

Download the ZIP of the skill from this page and unzip it. To make it available to every project, place the skill folder in ~/.claude/skills/. To limit it to one project, place it in that project's .claude/skills/ folder.

If you use the Claude apps instead, upload the skill under Settings → Capabilities → Skills.

  • Personal install: ~/.claude/skills/
  • Project install: .claude/skills/ inside your project
  • Claude apps: Settings → Capabilities → Skills, then upload

Check the requirements before you start

The skill needs an image generation backend. It can use Codex imagegen, Codex CLI with an active login, Cursor's GenerateImage tool, or a baoyu-image-gen installation. If none is available, the skill stops and asks how to proceed rather than drawing substitute SVG or HTML art.

The skill runs scripts and uses the network. Its metadata lists bun or npx as optional binaries; the script resolver prefers bun, then npx -y bun, and suggests brew install oven-sh/bun/bun if neither is present.

  • An image generation backend (Codex imagegen, Codex CLI with login, Cursor GenerateImage, or baoyu-image-gen)
  • bun or npx, for the bundled merge scripts
  • Network access for the image backend

Trigger the skill with natural requests

The skill loads when your request matches its description. Phrases from its description include "create slides", "make a presentation", "generate deck", "slide deck" and "PPT". Give the agent your source material and say what you want.

Expect a confirmation step. By default the agent proposes a plan and waits for your reply before generating any images. You can skip confirmation only by saying so explicitly in your request, for example "直接生成" or "不用确认".

  • "Make a slide deck from this article: [paste or path]"
  • "Create slides summarising these meeting notes for my team"
  • "Generate a deck in a clean, consistent style for a LinkedIn carousel"
  • "Make a PPT from notes.md"

Review the plan and the prompt files

Before images are rendered, the agent writes each slide's full prompt to a standalone file under prompts/, named NN-slide-[slug].md. These files are the reproducibility record: you can read them, reuse them, or switch image tools without regenerating the prompts.

Read the outline during the confirmation step. Ask for changes to style, audience, slide count or language there, since that is the cheapest point to change direction.

  • Prompt files live in prompts/ and follow the NN-slide-[slug].md pattern
  • Generation runs in batches; the default batch size is 4 unless you ask for another
  • Text inside images can be imperfect. Ask the agent to regenerate or simplify on-image text rather than editing the picture

Safety notes: scripts, network and sensitive content

The skill runs bundled TypeScript scripts with bun or npx and sends slide content to whichever image backend is used. Treat the image service as an external party. Do not send confidential or sensitive material through an external image service.

The skill's rules also forbid repairing a generated image by painting over its text with ImageMagick, Pillow, SVG, HTML or OCR scripts. Text problems should be fixed by regenerating from a corrected prompt or simplifying the on-image text.

  • Content is passed to the chosen image backend, so check that backend's data handling
  • Scripts run locally with bun or npx; read them first if your policy requires it
  • Generated text should be checked by you before sharing

What people use it for

  • Turning a long article into a shareable multi-slide explainer
  • Creating a social-media carousel from written notes
  • Producing a consistent-style set of slides across several topics
  • Drafting a visual summary of a meeting or report for colleagues

Frequently asked questions

Is the baoyu-slide-deck skill free to use?

The skill is published under the MIT licence. The source does not list prices for the image backends, so check the terms of whichever backend you connect, such as Codex or Cursor.

What do I need to install baoyu-slide-deck?

Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it in the Claude apps under Settings → Capabilities → Skills. You also need an image generation backend available to the agent.

Can it make editable PowerPoint slides?

No. The output is generated slide images, not editable slides. The skill can merge slides into PPTX or PDF with its bundled scripts, but the text inside images cannot be edited as PowerPoint text.

Why does the agent ask me to confirm before generating?

The skill's default is to confirm before generation. Explicit skill invocation or saved defaults are treated only as recommendations. You can skip confirmation by saying so explicitly in your request.

What if the text in my slides is wrong or unclear?

The skill does not paint over generated images. Ask the agent to regenerate from a corrected prompt or to simplify the on-image text.

What happens if I have no image generation tool?

The skill stops and tells you, then asks how to proceed. It does not substitute SVG or HTML drawings for raster images.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use this skill if you need a shareable slide deck or social carousel quickly from written content and already have an image generation backend. Pick a different tool if you need live presentations with editable PowerPoint text, or if your material is confidential and cannot go to an external image service.

Download the skill