Home / All Skills / Content & Media Creation / canghe-article-illustrator
English (EN)简体中文 (ZH)日本語 (JA)
Content & Media Creation · freestylefly/canghe-skills

canghe-article-illustrator

Analyzes article structure, identifies positions requiring visual aids, generates illustrations with Type × Style two-dimension approach. Use when user asks to "illustrate article", "add images", "generate images for article", or "为文章配图".

L4 High-Privilege · Graded auth & full auditV1 Commit-PinnedExecute CommandWrite FileDual-pass Audit
Author freestyleflyGitHub freestylefly/canghe-skills ↗License MIT（仓库根 LICENSE）Commit dd0bf35595
Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.

1How It Works & Mechanism

The core lies in the orthogonal two-dimensional Type (information structure) × Style (visual aesthetics) model, allowing flexible combinations of illustration schemes.

skills/canghe-article-illustrator/SKILL.md
| **Type** | Information structure | infographic, scene, flowchart, comparison, framework, timeline |
What is happening here: Decoupling "what images to provide" into two independent dimensions—Type determines the information structure (data/infographic, narrative/scene, process/flowchart…), while Style determines visual feel (notion/warm/blueprint…), both freely combinable.

The workflow is fixed into a six-step checklist, where "loading preferences" and "saving prompt files to disk before generation" act as hard blocking gates.

skills/canghe-article-illustrator/SKILL.md
- [ ] Step 1: Pre-check (EXTEND.md, references, config)
What is happening here: The six steps (Pre-check→Analyze→Confirm→Outline→Generate→Finalize) form the process backbone, with Step 1.5 and Step 5 explicitly marked as ⛔ BLOCKING.

Preferences rely on EXTEND.md, resolved with project-level > user-level priority; first-time setup must run if missing.

skills/canghe-article-illustrator/SKILL.md
test -f .canghe-skills/canghe-article-illustrator/EXTEND.md && echo "project"
What is happening here: Uses `test -f` to detect paths at both levels; if found, it parses and displays a summary; if not found, it enforces running first-time-setup.

The content analysis phase requires converting metaphors into "underlying concept visualisations", strictly forbidding literal illustrations.

skills/canghe-article-illustrator/SKILL.md
**CRITICAL**: Metaphors → visualize underlying concept, NOT literal image.
What is happening here: Following the four analysis tasks in Step 2 (content type/objective/core arguments/image positions), a CRITICAL boundary rule restricts metaphor visualization.

Confirmation is condensed into a single AskUserQuestion invocation (max 4 questions), with Type, Density, and Style being mandatory.

skills/canghe-article-illustrator/SKILL.md
**ONE AskUserQuestion, max 4 Qs. Q1-Q3 REQUIRED.**
What is happening here: Multiple-choice questions are asked all at once to prevent multi-turn back-and-forth; Q4 (language) appears only when the article language conflicts with EXTEND.md settings.

Structured prompt files must be persisted to disk prior to generation, and LABELS must incorporate actual article data.

skills/canghe-article-illustrator/SKILL.md
⛔ **BLOCKING: Prompt files MUST be saved before ANY image generation.**
What is happening here: Step 5 mandates writing `prompts/NN-{type}-{slug}.md` per template (ZONES/LABELS/COLORS/STYLE/ASPECT) for each illustration, prohibiting passing `--prompt` directly without saving to disk first.

Output directory and naming are standardized: 2-4 word kebab-case slug, appending timestamps on conflicts.

skills/canghe-article-illustrator/SKILL.md
**Slug**: 2-4 words, kebab-case. **Conflict**: append `-YYYYMMDD-HHMMSS`.
What is happening here: Artifacts are saved in `illustrations/{topic-slug}/`, containing source, `outline.md`, `prompts/`, and `NN-{type}-{slug}.png`.

Over 20 style definition files each detail background, hex palette, visual elements, Do/Don't, and Best For, compiled into styles.md.

skills/canghe-article-illustrator/references/styles/editorial.md
Magazine-style editorial infographic for professional content
What is happening here: Each style acts as an independent reference loaded on demand; editorial is positioned as "magazine-grade professional infographic" with a 9-role hex color palette.

2Core Capabilities

01Two-dimensional selection (Type × Style) freely combines illustration directions
02Six illustration types (infographic/scene/flowchart/comparison/framework/timeline)
0320+ visual style library (notion/warm/minimal/blueprint/watercolor…), each with built-in palettes and rules
04Density control (minimal/balanced/per-section/rich) determines image quantity
05Structured prompt templates (ZONES/LABELS/COLORS/STYLE/ASPECT) + mandatory disk-first persistence
06outline.md outline artifact (location/purpose/visual contents/filename)
07Modification workflow (Edit/Add/Delete actions, including prompt and outline updates)
08Preference persistence and output directory options (same-dir/illustrations-subdir/independent)

3External Dependencies

KindDependency / Endpoint
api图像生成 skill（由 Step 5.6 在宿主可用 skill 中挑选；SKILL.md 未点名具体后端）

4Recommended Governance & Evidence

Suggested Controls for Safe Operation
For Agent Runtime
Graded/per-use authorization, sandbox recommended
For Administrators
Approval workflow, rollback support, full audit
Evidence & Triggers (File Paths & Excerpts)
  • 本地工作目录: File I/O and artifact output: Agent writes to disk driven by SKILL.md instructions: source-{slug}, outline.md, prompts/*.md, and *.png under illustrations/{topic-slug}/
  • SKILL.md: Declares invocation of host CLI interpreters or external execution environments
Audit Notes: (a) The stated purpose of the skill is to generate illustrations from article content: Step 5.6 'Select generation skill, process references' directs the agent to hand off constructed prompts (containing real article data) to host-available image generation skills/backends, sending article contents and reference images outbound to third-party image models and consuming account credits. (b) Local disk writes are explicit: source, outline.md, prompts/*.md, *.png under illustrations/{topic-slug}/, and EXTEND.md. (c) The skill's directory contains zero scripts, zero built-in network endpoints, and zero credential reading; outbound traffic and credentials reside entirely within the invoked generation backend, thus avoiding a red rating. Although no specific backend is named (it states 'Select generation skill'), the workflow inherently transmits material to external generation services. Following precedents such as baoyu-article-illustrator and ai-image-generation, routing content to generation backends is uniformly marked orange.

5Second-Pass Verification

  • [ok] 自身目录无任何脚本或可执行文件 — 文件清单只有 .md；扫描“执行外部程序/写盘/网络请求”均为 0 命中，脚本执行面为空。
  • [discrepancy] 外部生成后端名称 — SKILL.md 仅写 'Select generation skill'，未像仓库内其它 skill 那样点名具体后端；出网事实由 skill 的声明用途（生成插画）推出，档位仍为 orange。
  • [ok] EXTEND.md 查找只探测存在性 — $HOME 命中仅是 test -f 的存在性判断，未读取文件内容之外的凭证或环境变量。
  • [ok] Type × Style 取值集合一致 — SKILL.md Types 表 6 项、Styles 表 6 项示例与 references/styles.md 汇总一致，未发现夸大。
  • [ok] prompt 先落盘纪律 — SKILL.md Step 5 明确 'Prompt files MUST be saved before ANY image generation'，并禁止未落盘直接 --prompt。
  • [ok] meta 元数据 — commit dd0bf355…、path skills/canghe-article-illustrator、license MIT 与 bundle §0 核定值一致。

6Verdict & Scenarios

  • Orthogonal two-dimensional selection makes "what to illustrate/how it looks" explicit and controllable, yielding far more predictable outcomes than generic single-prompt generation.
  • Strict reproducibility: mandates saving structured prompt files to disk prior to generation, with LABELS containing actual article data.
  • Rich style catalog: 20+ standalone style files with hex palettes and Do/Don't rules, loaded on demand without bloating context.
  • Clear semantic boundary: strictly requires visualizing underlying concepts rather than drawing metaphors literally.
    • Best for: Writers or content teams who need batch illustration generation for long articles/tutorials/technical posts by section, requiring visual consistency and reproducible prompts.
    Not for: Environments that are strictly offline or prohibit outbound data transmission; tasks seeking only a single cover image (use canghe-cover-image instead); fast draft generation unwilling to go through confirmation steps.
    Install Agent-Ready
    ① Verified Mirror 2026-10-10
    Option A · Direct DownloadDownload canghe-article-illustrator.tar.gz
    sha256: 0fd4e11132236e3d…
    Option B · Copy Guide for Agent
    install.json
    Agents automatically download and verify mirrors from this JSON guide.
    ② Upstream GitHub
    GitHub Repo ↗
    Pinned to commit dd0bf35595.