Guide · Agent skill
How to use the Article Illustrator agent skill (baoyu-article-illustrator)
Updated · written with AI from the project's own documentation and checked against our catalogue data
Article Illustrator is an agent skill that reads an article, finds the places where a picture would help, and proposes an illustration for each one. This guide is for bloggers, newsletter writers and content teams who want consistent images across a long piece. By the end you will know what the skill does, how to install it, how to ask for illustrations in plain language, and what it needs to run.
Article IllustratorJimLiu/baoyu-skills · MIT Download the skill Quick start
- 1Download and unzip. Download the baoyu-article-illustrator ZIP from this page and unzip it into ~/.claude/skills/ or your project's .claude/skills/ folder.
- 2Or upload in the Claude apps. In the Claude apps, open Settings, go to Capabilities, then Skills, and upload the skill.
- 3Confirm a backend is available. Make sure Codex imagegen, a codex login, Cursor GenerateImage or a baoyu-image-gen skill is available in your tool.
- 4Ask in plain language. Paste your article or give its file path and ask the agent to illustrate it or add images.
- 5Answer the setup questions. If no EXTEND.md exists, answer the first-time setup questions so your preferences are saved.
- 6Review and confirm. Check the proposed illustration positions, type, style and palette, then confirm before generation starts.
- 7Check the prompts and images. Review the prompt files in prompts/ and the generated images, and ask for regeneration where needed.
What is an agent skill, and what does Article Illustrator do?
An agent skill is a folder containing a SKILL.md file and supporting files that an AI agent loads when a request matches the skill's description. Article Illustrator is one such skill, published by JimLiu in the baoyu-skills collection under the MIT licence.
Once triggered, the agent analyzes the article's structure and identifies positions that need a visual aid. For each position it builds the illustration using a three-dimension approach: Type (information structure, such as infographic, scene, flowchart, comparison, framework or timeline), Style (rendering approach) and Palette (colour scheme). Keeping these fixed across the run is what makes the images look like a set.
- Analyzes article structure and finds illustration positions
- Applies one Type × Style × Palette system across all images
- Saves each image's full prompt to a file under prompts/ before rendering
- Asks you to confirm before any image is generated
Install the skill (ZIP download or Claude apps upload)
Download the baoyu-article-illustrator ZIP from this page and unzip it. To make it available to Claude Code for all your projects, place the 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, open Settings, go to Capabilities, then Skills, and upload the skill there.
- Personal install: ~/.claude/skills/
- Project install: the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills
Set up an image generation backend
The skill does not draw images itself. It sends each saved prompt to an image backend that is available in your tool. The skill checks in this order: a backend you name in the current message, then the preferred_image_backend setting in an optional EXTEND.md file, then auto-selection.
Auto-selection prefers Codex's native imagegen skill when running inside Codex, then Cursor's GenerateImage tool, then other runtime-native image tools, then a single installed non-native backend such as baoyu-image-gen. Codex can also be reached through the codex CLI when it is on PATH with an active codex login.
If no backend is available, the skill tells you and asks how to proceed. It will not substitute SVG, HTML or canvas drawings.
- Codex imagegen (native skill inside Codex)
- codex CLI on PATH with an active codex login
- Cursor GenerateImage tool
- A baoyu-image-gen skill
- Optional EXTEND.md preferences file for saved defaults
Trigger the skill with a natural request
You do not need a special command. Ask in plain language, and paste the article or give its file path. The skill's description lists trigger phrases such as "illustrate article", "add images", "generate images for article" and "为文章配图".
On first use, if no EXTEND.md is found, the skill runs a first-time setup and asks its questions before continuing. Then it analyzes the article and proposes illustration positions and options for you to confirm.
- "Illustrate this article: [paste text or file path]"
- "Add images to my newsletter draft in docs/post.md"
- "Generate images for this article in the same house style"
- "为文章配图" followed by the article
Review the plan and confirm before generation
By default the skill confirms before generating. Your explicit invocation, file path, presets and EXTEND.md defaults only recommend options; they do not skip confirmation. Generation starts only after you answer the confirmation step.
To skip confirmation, say so in the request, for example "直接生成", "不用确认" or "跳过确认". The skill will then state the assumed type, density, style, palette, language and backend before generating.
- Check the proposed positions and number of images
- Check the chosen type, style and palette
- Reply to confirm, or ask for changes
How images are generated and checked
Each image's full, final prompt is written to a standalone file in prompts/, named NN-{type}-[slug].md, before any backend is called. This record lets you reproduce or re-run an image later, or switch backends without rewriting prompts.
Images are generated in batches. If the backend has a native batch interface, it is used. Otherwise the skill can dispatch up to generation_batch_size images at a time, with a default of 4. Failed items are retried once without regenerating successful ones.
The skill never repairs text by painting over a generated image. If text in an image is wrong, it regenerates from a corrected prompt, redraws with less text, or asks which candidate to keep.
- Prompt files are saved under prompts/ before rendering
- Batches default to 4 images, overridable in the request
- Reference images can be supplied with --ref or pasted in conversation
What people use it for
- Adding a set of consistent infographic-style images to a long blog post
- Illustrating a newsletter draft in one house style across several editions
- Producing a series of article images for a content team using the same palette
- Creating timeline or flowchart-style visuals for process-heavy articles
Frequently asked questions
Is Article Illustrator free?
The skill is published under the MIT licence, as stated on this page. Image generation depends on the backend you use, such as Codex, Cursor or baoyu-image-gen, and those may have their own terms.
What do I need to use baoyu-article-illustrator?
You need an image generation backend: Codex imagegen, the codex CLI with an active login, Cursor's GenerateImage tool, or a baoyu-image-gen skill. An optional EXTEND.md file can save your preferences.
Does the skill run scripts or use the network?
According to this listing, the skill runs no scripts of its own but does use the network, because image generation is performed by a backend. Review the backend you connect before use.
How do I install it in Claude Code?
Unzip the downloaded folder into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder to limit it to that project.
Can it add text to images accurately?
Not reliably. The skill forbids editing text on generated images and relies on regeneration, which may still be imperfect. Avoid it for images needing exact labels, charts or data.
What happens if I have no image tool set up?
The skill stops and asks how to proceed. It will not draw SVG or HTML substitutes, so you need a raster image backend installed.
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
Article Illustrator suits bloggers, newsletter writers and content teams who want a consistent set of illustrations for long articles and already have an image backend such as Codex, Cursor or baoyu-image-gen. It is a poor fit if you need exact text, charts or data graphics inside images, or if your project forbids AI-generated images. In those cases, choose a diagramming or design tool instead.
Download the skill