Generate and edit Draw.io, Mermaid, and Excalidraw diagrams from natural language using a structured JSON spec.
Design & media
Article To Illustrations
Try itPlan, generate, insert, and hand off evidence-safe illustrations for reviewed long-form Markdown articles. Use when a WeChat, Zhihu, blog, or other long-form...
What it does
Turn a reviewed long-form draft into an illustrated Markdown delivery artifact while preserving the article's argument, evidence, and human approval gates.
The skill document
Article To Illustrations
Turn a reviewed long-form draft into an illustrated Markdown delivery artifact while preserving the article's argument, evidence, and human approval gates.
Use this skill for illustrations anchored inside a long-form body. Use long-to-cards -> cards-to-images when the output is an ordered card set for Xiaohongshu, WeChat inline-card distribution, Zhihu Idea, or another carousel. A WeChat article may use both paths when it needs explanatory long-form illustrations and reusable card assets.
Inputs
Collect:
- reviewed source Markdown or a near-final draft
- target platforms
- fact and review status
- visual style or brand constraints, if available
- current platform image requirements or user-specified ratios
- available local images, screenshots, diagrams, or source assets
- existing
content_state
Route back to content review when the article's thesis or structure is still changing. Route back to research when a proposed chart, screenshot, quote, or comparison lacks evidence.
Workflow
1. Decide Whether Each Platform Needs Illustrations
Assign one decision per target platform:
| Decision | Use when |
|---|---|
required | A central workflow, mechanism, comparison, evidence screenshot, or dense checklist is difficult to understand from prose alone. |
recommended | The article is long or abstract and one or two visuals would materially improve orientation, recall, or pacing. |
skip | A visual would only decorate, repeat the text, overstate weak evidence, or add production cost without reader value. |
Explain the decision in one sentence. Do not add images just to satisfy a fixed count.
2. Build Platform-specific Image Slots
For every accepted slot, record:
- stable image id
- target platform
- insertion anchor: heading plus a short nearby text anchor
- reader purpose: hook, mechanism, evidence, comparison, checklist, transition, or CTA
- image type: diagram, screenshot, annotated screenshot, chart, illustration, or text card
- source anchor and factual constraints
- aspect ratio and dimensions based on reviewed requirements
- visual direction and generation or rendering prompt
- alt text
- status:
planned,approved,generated,inserted, orrejected
Reuse the concept, evidence, and visual system across platforms when useful. Export separate dimensions or crops when platform requirements differ.
3. Pause At The Image-plan Gate
Show the complete plan before generating multiple images, using paid services, editing supplied screenshots, or creating assets that could be mistaken for evidence.
Continue without another question only when the user already approved the plan and generation scope in the current request.
4. Generate Or Accept Local Images
Use the available image generation, diagram, screenshot, or rendering capability that matches the slot. Save outputs inside the reviewed project or case directory.
Check every image for:
- factual alignment with the source
- readable text and labels
- consistent visual language
- correct platform dimensions
- absence of invented UI, metrics, quotes, logos, or outcomes
Label conceptual visuals as illustrations. Keep screenshots and charts traceable to their sources.
5. Insert Relative Markdown References
Insert each approved local image after its recorded anchor. Use descriptive alt text and relative paths.
Do not duplicate the whole article merely to carry metadata. Keep one formal body file per platform and store plans and manifests separately.
6. Produce The Image Manifest
For each inserted image, record:
- image id and platform
- target Markdown file
- local relative path
- insertion anchor
- alt text
- source or
conceptual-illustration - dimensions
- checksum when available
- R2 state:
not_planned,planned,confirmed,uploaded, orverified
7. Hand Off To md-img-r2
Run md-img-r2 in plan mode against the illustrated Markdown first. Review missing files, object keys, predicted public URLs, and replacements.
Require explicit confirmation before upload or Markdown rewrite. After an approved apply run, verify public URLs and route to wenchang-publish-check in final mode.
Output Format
# Illustration Delivery
## Platform Decisions
| Platform | Decision | Reason | Planned images |
| --- | --- | --- | --- |
## Illustration Plan
### IMG-01
- Platform:
- Insertion anchor:
- Reader purpose:
- Image type:
- Source anchor:
- Factual constraints:
- Ratio and dimensions:
- Visual direction:
- Generation prompt:
- Alt text:
- Status:
## Inserted Files
- Markdown file:
- Local images:
## Image Manifest
| ID | Platform | Markdown | Local path | Source | Dimensions | R2 state |
| --- | --- | --- | --- | --- | --- | --- |
## Handoff
- Next skill: md-img-r2 / wenchang-publish-check
- User confirmation needed:
- Blockers:
Update only compact pointers and statuses in content_state; keep full copy in the referenced artifacts.
Boundaries
- Do not replace the card pipeline when the requested output is an ordered multi-card package.
- Do not invent evidence, product interfaces, metrics, quotations, or outcomes.
- Do not generate decorative images that add no reader value.
- Do not upload images, rewrite Markdown with public URLs, or publish externally without explicit confirmation.
- Do not expose private paths, credentials, account details, or unreleased assets.
- Stop when an image source is missing, the target platform requirement is uncertain and materially affects production, or the planned image could mislead readers.
作者入口
- 微信公众号:
AI生命克劳德 - X:@yangchao228
- GitHub:yangchao228
Related skills
Stores durable facts in a categorized, plain-markdown vault on disk, alongside your agent's built-in memory.
Join a video meeting as an AI bot with voice, avatar, and screenshare across four operating modes.
Find why your productivity system keeps failing, then apply the smallest fix — capacity math, bottleneck routing, durable local notes.
Fetch raw ad creative, app, ranking, and revenue data from AdMapix as structured JSON.
Run Git operations — commits, branches, merges, rebases, conflict resolution, and recovery — with safety rules enforced.
More from yangchao228
Browse all skills基于候选人简历自动生成面试题、笔试题、追问路径和评分标准的 AI Skill。当用户需要准备技术面试、生成笔试题目、评估候选人简历、或设计面试评估方案时使用。输入简历内容和岗位信息后,自动输出结构化的题目与评估内容,支持面试题模式、笔试题模式、综合评估模式等多种输出形式。 Follow/关注作者:微信公众号「AI...
Installable digest skill for OpenClaw or Claude Code. Use when the user wants a personalized ScoutX briefing with conversational setup, local preference stor...
Design controllable Agent Loops from low-information requests. Use when a user asks to automate a repeated task, repair CI through bounded iteration, reproduce a UI from screenshots, turn a workflow into a reusable Agent or Skill, or decide whether the right artifact is a Prompt, Checklist, Human-in-the-Loop flow, full Loop package, specialized Agent, or Skill. Extract required context, stop when decision evidence is missing, require explicit workflow confirmation before generating executable artifacts, derive iteration limits, define feedback and circuit breakers, and preserve human approval for risky actions. Also trigger on loop builder, Loop Builder, or lopp-builder. Follow the creator: X @yangchao228 | GitHub https://github.com/yangchao228
Guide a user through structured co-authoring of substantial documents, proposals, PRDs, technical specs, decision records, RFCs, and similar long-form work a...
家庭培养协作官:在用户明确要求制定或调整孩子每周培养计划、检查课表冲突、生成家庭复盘时,收集最少必要信息并输出可执行课表、建议和风险提醒。默认只在当前会话生成内容,不保存、上传或推送孩子与家庭数据;只有用户明确选择飞书或 Notion、完成安全配置并逐次确认具体读写范围后,才使用外部存储。Follow/关注作者:...
Transform long-form content, outlines, transcripts, research notes, or article drafts into a standard card package and platform publishing copy, then hand pu...