Book Illustration Workflow: Screenshots to Feishu, Automated

Where this fits

You're mid-way through writing a technical book. Chapter 7 needs screenshots that show a real tool workflow, but you can't stand manually listing prompts, naming files, placing images in Markdown, cleaning out your author notes, and then re-doing the ordering in Feishu. This skill abstracts that whole process into a repeatable, step-driven workflow for coding agents.

Why agents benefit

• Agents can turn a request like 'handle the illustrations for this chapter' into a concrete plan: required screenshots, exact Claude Code prompts, and file names tied to figure numbers.
• Agents enforce a clean separation: test data and demo scaffolding are prepared outside screenshots, so the captured content looks like real work, not staged examples.
• Agents locate the right insertion point by reading the prose, not by dumping all images into a placeholder block, preserving semantic order.
• Agents can strip author-only remarks from a local Markdown draft to produce reader-ready copy.
• Agents can sync the cleaned chapter and images to Feishu with the same order and file mapping.

Practical scenarios

Moving from vague request to step-by-step execution

The author says, 'Do the screenshots for Chapter 5.' The skill produces a sequence: send this prompt to Claude Code, capture this screen, save as 05-3-config.png, then place it after the sentence about configuration keys. Each step is one action, so the author stays in control.

Cleaning Markdown before publishing

The draft contains notes like 'insert screenshot here' and 'this example is simplified for the book.' The agent distinguishes reader-facing text from editorial debris, removes the notes, and places images where the text actually references them.

Syncing to Feishu in the right order

Instead of uploading images as a heap, the agent follows the chapter's figure-number mapping, uploads each image at the correct point in the Feishu document, and preserves the cleaned text order.

Add it to your agent workflow

Install the skill:

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

Then ask something like: 'Use the book illustration workflow for Chapter 7. Tell me the first three screenshots I need.' The agent will respond with the exact prompt for Claude Code, the screen to capture, the filename, and where it belongs in the Markdown. You confirm, and it moves to the next step.

Compare before adopting

Compare before adopting:

• Quality signals: Does the skill enforce the 'no book words in screenshots' rule and semantic placement? Can it clean author notes without touching code blocks?
• Maintenance freshness: When was the last commit? Are the Feishu-sync instructions still aligned with current API behavior?
• Alternatives: You could build this workflow as your own prompt template, but you'd lose the built-in checks and file-naming conventions.
• Workflow fit: Does your book project use local Markdown and Feishu? This skill assumes both.

Why it is worth tracking

The skill is a focused example of a process skill for long-form content production: it turns recurring editorial chores into agent steps, and its 566 GitHub stars indicate real usage. Monitor it when your own book workflow involves tight coupling between prose, code snippets, and output documents. If the author adds more platform integrations or refines the naming rules, that's a sign the process is actively evolving. For more details and updates, you can check its listing at https://www.openagentskill.com/skills/kangarooking-book-illustration-workflow.