An open, identity-neutral edition of the Xuemu Lab hand-drawn explainer workflow. It turns a technical article or concept into a structured shot plan, reproducible image prompt, and visual QA gate.
This public package contains no Xuemu character library, no private identity assets, no customer data, no browser validation records, and no third-party artwork.
git clone https://github.com/aierlanjiu/xuemu-illustration-core.git
cd xuemu-illustration-core
./install.shRefuses to overwrite existing skills. To preserve an existing installation:
./install.sh --backup-existingThe skill strictly enforces a four-stage progression from input concept to verified visual output:
Formulate a single, clear one-sentence communication goal for the illustration.
Select a composition pattern (e.g. choice and judgment) and map subjects, labels, and reading path.
Construct a structured image prompt covering composition, line weight, muted colors, and negative constraints.
Verify readability at mobile size, text limits, aspect ratio, and identity boundaries before finalizing.
The SVG composition below demonstrates the choice and judgment layout pattern from examples/placeholder-output.svg.
| Path | Role |
|---|---|
| SKILL.md | Core agent definition, boundaries, defaults, and workflow contract |
| install.sh | Safe installer script with backup flags |
| references/composition-patterns.md | 5 foundational layout patterns for technical explainers |
| references/shot-config-template.md | Structured template for shot planning |
| references/qa-checklist.md | Acceptance criteria for legibility, style, and boundaries |
| examples/placeholder-output.svg | Original layout composition reference |
| ASSET_LICENSE.md | Explicit asset boundary and brand distinction policy |
| SECURITY.md | Data handling rules and secret prevention guidelines |
View source code, report security issues, or clone the public package on GitHub.
Open xuemu-illustration-core on GitHub