book-illustration-workflow

Original：🇨🇳 Chinese

Translated

This skill handles the workflow of chapter screenshots and illustrations during book writing. It applies to: sorting out which screenshots are needed for a chapter, providing step-by-step practical prompts for Claude Code, defining the mapping between screenshot filenames and figure numbers, filling image positions in local Markdown, cleaning up author notes to create reader-facing text, and synchronizing chapters and images to Feishu Docs in the correct positions. This skill should be triggered when users mention terms like "book screenshots", "chapter illustrations", "figure number correspondence", "insert into original text", "upload to Feishu Docs", or "follow the previous workflow".

9installs

Sourcekangarooking/kangarooking-skills

Added on2026-05-02

NPX Install

npx skill4agent add kangarooking/kangarooking-skills book-illustration-workflow

SKILL.md Content (Chinese)

View Translation Comparison →

Book Illustration Workflow

This skill handles the complete workflow of "chapter screenshots from planning to insertion into text and synchronization to Feishu".

Objectives

Break down the illustration workflow for a book chapter into a stable process:

Identify the screenshots and final images actually needed in the chapter
Provide real, executable prompts for Claude Code step by step
Ensure one-to-one correspondence between screenshot filenames, figure numbers, and positions in the text
Place images back in the correct positions in local Markdown
Remove author notes that are not intended for readers
Synchronize the reader-facing chapter and images to Feishu

Trigger Conditions

Use this skill when users have any of the following intentions:

Ask you to handle the screenshot workflow for a book chapter
Ask you to provide step-by-step guidance for taking screenshots or generating images with Claude Code
Ask you to organize the mapping of "figure number -> prompt -> filename -> position in original text"
Ask you to insert images back into local articles
Ask you to synchronize chapters to Feishu Docs
Ask you to "continue with the previous workflow"

Core Principles

1. Prompts in screenshots must resemble real work tasks

The following terms are prohibited in prompts for Claude Code:

```
book
```
```
chapter
```
```
case
```
```
demo
```
```
screenshot
```
```
suitable for display
```
```
chapter7-demo
```
```
for book writing
```

User input in screenshots must look like real office tasks.

2. Data preparation should be done outside screenshots

Do not include "creating test data", "setting up directories for screenshots", or "preparing samples for display" in Claude Code conversations.

Correct approach:

Prepare data outside screenshots
Only show real usage within screenshots

3. Only advance one step at a time

If the user requests "let's take it step by step", only provide the following each time:

Content to send to Claude Code
Which screen to capture
Screenshot filename
Corresponding position in the original text
Whether there is already a placeholder in the original text

Provide the next step only after the user completes the current one.

4. Place images according to text semantics, not placeholder blocks

If the article contains editing traces like:

```
material placeholder
```
```
illustration placeholder
```
A group of figure numbers clustered at the start of a section

Do not mechanically insert all images into that block.

Correct approach:

Find the sentence in the text that actually introduces the image
Place the image right after that sentence or paragraph
Follow this order when uploading to Feishu as well

5. Reader-facing text must not retain author notes

The following content should not appear in reader-facing text or Feishu Docs:

```
material placeholder
```
```
illustration placeholder
```
```
suggested shooting method
```
```
suggested approach
```
```
illustration description copy
```
```
source hint
```
```
If we add an image here...
```
```
If you include an image in the book...
```
```
You can... in the book
```

Processing principles:

Rewrite valuable information into reader-facing text
Delete pure editing notes directly
Replace completed screenshot placeholders with actual images

Standard Workflow

Phase A: Chapter Sorting

Locate the target chapter Markdown file
Identify all figure numbers, code blocks, table placeholders, and "material placeholder" sections
Classify each image type:
- Raw data screenshots
- Claude Code conversation screenshots
- Final charts/reports
- External official page screenshots
- Post-edited composite images
Provide the minimum executable screenshot sequence

Priority order is usually:

Installation or environment confirmation
Raw data images
Core task invocation images
Result images
Comparison images
Extended application images

If you need to quickly create a full chapter screenshot plan, prefer the master table template in [references/chapter-illustration-template.md]. It is suitable for listing "figure number, purpose, prompt, filename, position in text, completion status" all at once before official screenshotting.

Phase B: Step-by-Step Screenshot Guidance

If the user wants to execute step by step, output using this template:

md

Step N: <Action Name>

Content to send to Claude Code:
```text
<Real Prompt>

Which screen to capture:

<Clearly specify to retain input / output / result image>

Screenshot filename:

<filename.png>

Corresponding position in original text:

<Position in chapter or figure number description>

Is there a placeholder in the original text now:

Yes / No


Requirements:

- Do not jump to subsequent steps
- Do not provide too much information at once
- Filenames should be directly usable for saving

### Phase C: Fill Back into Local Article

First check which type the local Markdown belongs to:

1. `Reader-facing text + independent image placeholders`
2. `Author-facing text + large material placeholders`
3. `Text with numerous embedded editing notes`

Processing methods:

- If it is still the author version, first clean it into the reader version
- If images are ready, directly insert Markdown image syntax into the text
- If images are not ready yet, only keep concise reader-friendly expressions and remove editing notes

Preferred image insertion format:

```md
![Figure 7-10 Example of Completed Trend Chart](Chapter7_Images/Figure7-10_Completed_Trend_Chart.png)

Phase D: Synchronize to Feishu

When synchronizing to Feishu, do not blindly modify a large number of misplaced images in old documents. Preferred approach:

First revise the local Markdown into the reader version
Split the content in the order of "Markdown paragraph -> corresponding image -> next paragraph"
Create a new Feishu document and upload the content
Set the new document as the main document

Only modify the old document if the user explicitly requests to retain it.

Phase E: Feishu Upload Rules

When using

lark-cli

:

```
docs +create
```
to create a document
```
docs +update --mode append
```
to append markdown
```
docs +media-insert
```
to insert images

Notes:

Images must be inserted in the order they appear in the text
Do not insert images by "figure number order", strictly follow the order in the article
If there is a prefix conflict in filenames (e.g.,
```
Figure7-1
```
mistakenly matches
```
Figure7-10
```
), use exact matching

Exact matching rules:

Only match boundaries like
```
^(figure number)(_|\\.|$)
```
Do not use simple prefix inclusion

Naming Rules

Default screenshot filename format:

```
figure_number_brief_description.png
```

Examples:

Figure7-2A_Add_Official_Skills_Market.png

Figure7-8A_Call_xlsx_Data_Cleaning_Interface.png

```
Figure7-18_Report_Homepage_Example.png
```

Requirements:

Figure number must be at the beginning of the filename
Chinese descriptions should be as concise as possible
This naming is relied on for subsequent text insertion and Feishu upload

Attached Templates

Chapter Screenshot Master Table Template: See references/chapter-illustration-template.md

Quality Check List

Before declaring completion, you must self-check:

Whether each image matches its corresponding figure number
Whether prompts resemble real office tasks
Whether any author notes remain
Whether image order in local Markdown is correct
Whether images in Feishu follow the correct paragraphs
Whether there are issues like
```
Figure7-1
```
mistakenly matching
```
Figure7-10
```

Default Response Style

If the user says "follow this workflow", execute the following by default:

First check the target chapter file and image directory
If screenshot guidance is needed, enter step-by-step mode
If screenshots are already completed, prioritize revising the local Markdown
After confirming locally, synchronize to Feishu

Things Not to Do

Do not leave author's personal editing notes in reader-facing text
Do not reveal "this is for book writing" in screenshot prompts
Do not blindly insert images into large material placeholders
Do not repeatedly modify old Feishu documents randomly, causing more misalignment
Do not use vague file matching rules to find images in batches