From topic selection, writing, and image matching to quality inspection, it covers the complete creation workflow for operations articles. You only need to provide the material—an article link, a Skill, or a finished body of text—and leave the rest to it.
See the core execution workflow for details
references/core_workflow.md
Overview
The graphic-and-text operations content creator solves the most common efficiency bottleneck for content teams:Writing and illustrating an article often requires collaboration among multiple people and repeated polishing。
With this tool, you can:
- See a good article and directly imitate it into your own version, with consistent style but entirely new content
- Give it a Skill or product and automatically generate a professional promotional long-form article with images in place
- Finish the body text but not want to find images, drop it in and it automatically determines where images should go and what style to use
Three scenarios, fully covered by one tool.
Target users
- 📝 Content operations — quickly produce benchmark-style articles and product promotion copy, handling everything from writing to image matching
- 🛍️ Brand / skill developers — automatically generate promotional long-form articles aligned with product capabilities, so content and features stay connected
- 🏢 Bloggers / editors — intelligently add images to existing body text, saving the trouble of manually planning image placement
Features
🎯 Core features
- 🔗 Link imitation: Provide any article link, automatically capture the original content and images, learn its style and structure, and imitate it into a brand-new article. The image style stays consistent with the original, making it suitable for benchmark analysis and competitor content study.
- 📣 Skill promotion article: Provide a Skill file path or name, automatically parse the product capabilities and capture the latest information from the target site, and generate a promotional long-form article aligned with the product's actual capabilities. Images use the red fox explainer style, matching both visually and in content.
- 🖼️ Body text image supplementation: Provide complete body text, automatically analyze the article structure by section semantics and word-count density, intelligently decide image placement and quantity, and generate images with a consistent style in batches. Suitable for scenarios where you have finished the body text and do not want to manually find images.
- ✅ Automatic quality check: After each article is generated, automatically perform 9 quality checks (title length, number of sections, number of tables, word count, number of FAQs, prohibited words, paragraph line count, information density, etc.) and output a JSON-format quality inspection report.
Writing quality control
It is not about casually writing an article and calling it done. The article generation process has a complete built-in quality control system:
- Automatic article type matching: Automatically identify the topic as tool recommendation, methodology, or crisis response, with each type having its own writing focus and structural suggestions
- Reader persona adaptation: Supports two reader positions, individual creators and teams/enterprises, and automatically adjusts the emphasis (personally actionable methods vs. tool-system collaboration)
- Selectable tone: Direct and candid / professional and rigorous / light conversation, defaulting to direct and candid
- Controllable word count: Supports custom word count ranges (default 2000-3000 words); link imitation adapts to the original article's length
Automatic image matching capability
Article images are not simply "inserting a picture," but part of the content:
- Style transfer: In link imitation mode, the original image serves as a style reference, new content is automatically generated, and the visual style remains consistent
- Semantic matching: In Skill promotion mode, images automatically match the presenter's posture (welcome, analysis, thinking, operation, etc.) according to chapter semantics
- Smart chapter splitting: In body text image supplementation mode, image generation points are automatically decided based on H2 chapter segmentation + word count density, skipping positions unsuitable for images such as FAQ / conclusion / introduction
- Controllable quantity: The upper limit of images can be adjusted from 0-5, or the system can decide automatically
One-click installation
Prerequisites
- Python 3.8+ (used to run image generation and quality check scripts)
REDFOX_API_KEY(used for image generation; if not configured, text-only articles can still be generated)
Installation steps
- Add this Skill directory to your Agent skill list
- Obtain
REDFOX_API_KEY: Go to Redfox hub to register an account, and on the API Keys page create a key - Configure environment variables:
# macOS / Linux
export REDFOX_API_KEY="your_key_here"
# Windows PowerShell
$env:REDFOX_API_KEY="your_key_here"
- Verify installation: tell the Agent "help me write a promotional article for this Skill" and confirm the skill responds normally
Environment variable description
| Variable name | Required | Description |
|---|---|---|
REDFOX_API_KEY |
Conditionally required | Image generation service key; if not configured, articles are generated normally, and images are downgraded to placeholders or skipped |
Usage guide
Basic usage
1. Link imitation
Tell the Agent the link of the article you want to imitate:
User: Help me imitate this article, keep the style but switch it to my product https://example.com/some-article
Agent: The original article has been fetched, learning style and structure... The article has been generated, including 3 images, saved as article.md.
2. Skill promotional article
Provide the Skill file or name:
User: Help me write a promotional article for the Skill xhs-ranking
Agent: The Skill capabilities have been parsed and the target site information has been fetched. The promotional article has been generated, including 3 Redfox-style images.
3. Body text image supplementation
Paste the body text or provide the file path:
User: Help me add images to this body text (paste body text content)
Agent: The body text is 1800 words in total, 4 image positions have been identified, images are being generated... Completed, saved as article.md.
Advanced usage
- Specify tone: "Write in a professional and rigorous tone"
- Specify readers: "The target readers are team enterprises"
- Specify the number of images: "At most 2 images"
- Turn off images: "No images needed" or "set style to none"
- Specify word count: "Keep it around 1500 words"
Quick reference for common phrases
| Intent | Example phrasing | Effect |
|---|---|---|
| Link imitation | "Help me imitate this article, keep the style but switch it to my product" | Learn the style and structure of the source text to generate a brand-new article with matching illustrations |
| Skill Promotion | "Help me write a promotional article for this Skill" | Automatically generate a long promotional article and illustrations aligned with product capabilities |
| Add Illustrations to Body Text | "Help me add illustrations to this body text" | Intelligently analyze the article structure, automatically determine illustration placement and generate them |
| Competitive Analysis | "Analyze the writing style of this article and help me write one from a different angle" | Learn the structure and style of excellent content to produce differentiated content |
| Control Tone | "Write in a relaxed conversational tone" | Generate articles in a specified tone |
| Disable Illustrations | "No illustrations needed" | Output plain text articles only |
Use Cases
| Scenario | Role | Example Prompts | Benefits |
|---|---|---|---|
| Benchmark Imitation | Content Operations | "Help me imitate this article, keep the style but switch to my product" | Quickly produce high-quality articles with consistent style, saving the effort of starting from scratch |
| Skill Promotion | Skill Developer | "Help me write a promotional article for this Skill" | Automatically generate promotional articles aligned with product capabilities, so content and features stay connected |
| Article Illustrations | Blogger/Editor | "Help me add illustrations to this body text" | Intelligently select illustration placement and quantity, with no manual planning needed |
| Competitive Analysis | Brand Operations | "Analyze the writing style of this article and help me write one from a different angle" | Learn the structure and style of excellent content to produce differentiated content |
Output Description
After generation is complete, you will receive:
- Complete Article: saved as
article.md, including title, body text, FAQ (if any), and illustrations (embedded into the corresponding sections as) - Quality Check Report：
validation_report.json, including the results of 9 article quality checks - Full Text in Conversation: output the full article directly in the conversation
- Feedback Points: if you need to adjust a section or illustration, you can specify the direction of modification
FAQ
Q1: Can it be used without an API Key?
A: Yes. Without REDFOX_API_KEY , the article body is generated normally, and the image part is handled with degradation: link imitation uses placeholders instead, while Skill Promotion and Add Illustrations to Body Text skip illustrations and output plain text only.
Q2: Can the generated article be adjusted manually?
A: Yes. After generation, simply tell the Agent the direction you want to adjust (such as "make the third paragraph shorter" or "replace an illustration"), and the Agent will make targeted modifications based on the existing article.
Q3: Can the illustration style be customized?
A: Three styles are supported:redfox(Red Fox Explainer),reference(follow the original image style),none(no style modification). Just tell the Agent your preference.
Project Architecture
visual-ops-writer/
├── SKILL.md # 技能入口文件
├── references/
│ ├── core_workflow.md # 核心执行流程（Agent 执行链）
│ ├── writing-framework.md # 写作框架参考
│ ├── article-type-templates.md # 文章类型模板（工具推介/方法论/危机应对）
│ ├── persona-matrix.md # 语气×读者×视觉 IP 配置矩阵
│ ├── image-decision-rules.md # 配图决策规则
│ ├── link-rewrite-guide.md # 链接仿写指南
│ ├── prompt-templates.md # 提示词模板
│ └── error-handling.md # 错误处理
├── scripts/
│ ├── fetch_article.py # 文章抓取与解析
│ ├── validate_article.py # 机器化质量检查
│ ├── generate_image.py # 图片生成（支持文生图与风格迁移）
│ └── decide_images.py # 配图位置与数量决策
└── assets/
├── 02.jpg # 素材图片
└── 03.jpg # 素材图片