CommunityImagegithub.com

mascot-illustrations

Image skill. Generate 16:9 hand-drawn conceptual illustrations, warm watercolor or mixed-media studio scenes, and paired 3:4 + 4:3 Chibi watercolor dual-cover posters starring a configurable creator mascot IP (`{{IP_PROFILE}}`, default solid-ink studio mascot with #2EE59D accent), complete with calligraphic brand signatures, dynamic C/S-spine or 2.5-head Chibi poses, hand-drawn social/GitHub icons, authentic AI product wireframes (Cursor, Claude Code, ChatGPT Canvas, DeepSeek-R1, MCP, Doubao, Muse, Grok Bot,

What is mascot-illustrations?

mascot-illustrations is a Claude Code agent skill that image skill. Generate 16:9 hand-drawn conceptual illustrations, warm watercolor or mixed-media studio scenes, and paired 3:4 + 4:3 Chibi watercolor dual-cover posters starring a configurable creator mascot IP (`{{IP_PROFILE}}`, default solid-ink studio mascot with #2EE59D accent), complete with calligraphic brand signatures, dynamic C/S-spine or 2.5-head Chibi poses, hand-drawn social/GitHub icons, authentic AI product wireframes (Cursor, Claude Code, ChatGPT Canvas, DeepSeek-R1, MCP, Doubao, Muse, Grok Bot,.

Works with✓Claude Code✓Codex CLI✓Cursor✓Gemini CLI
npx skills add https://github.com/xeronsh/handdrawn-video-skills/tree/HEAD/skills/mascot-illustrations

Installed? Explore more Image skills: steipete/songsee, affaan-m/frontend-design, mattpocock/improve-codebase-architecture · View all 6 →

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Mascot Illustrations

Turn abstract judgments, workflows, AI tool comparisons, AI developer slang ("AI 黑话"), and video cover hooks into memorable hand-drawn ink & watercolor illustrations starring your configurable creator mascot IP ({{IP_PROFILE}}).

Portability Note: On Gemini Web Tasks, call image_gen:generate_image (with image_references for consistency). On Claude Code / Codex / local CLI, pass the rendered prompts to your configured image-generation tool (Imagen / Gemini Image API / GPT Image / Flux).

When to Use

  • Generating paired video cover posters (3:4 vertical + 4:3 horizontal for Douyin/Xiaohongshu, or 16:9 for Bilibili/YouTube)
  • Generating 16:9 body illustrations for articles, blog posts, technical docs, or methodology notes
  • Illustrating AI workflows, agent comparisons (Cursor, Claude Code, ChatGPT Canvas, DeepSeek-R1, MCP, 豆包, Meta Muse, Grok Bot, Agent CLI), AI engineering methods (Prompt Caching, Context Compaction, RAG, Harness), or AI slang metaphors (Vibe Coding, 抽卡式编程, AI 幻觉, Context Rot, Token 刺客, 炼丹蒸馏)
  • Producing warm watercolor scenes or mixed-media real-object studio scenes when tactile depth is requested

Steps

1. Load Profile & Select One of the 3 Explicitly Separated Canvas Modes

Load references/ip-and-style-dna.md and references/prompt-templates-and-qa.md (via get_skill_resources or direct file read) and populate {{IP_PROFILE}} slots from templates/ip-profile.example.yaml (default example: Ink Mascot / 墨水吉祥物, {{BRAND_NAME}}="InkCraft", {{ACCENT_HEX}}="#2EE59D"):

  • Mode 1.0 — Pure-White Editorial Illustration (aspect_ratio="16:9"): Pure #FFFFFF canvas edge-to-edge, 40%–55% drawn density, zero top-left title, 3–4 short handwritten labels (2–5 chars), crisp non-chibi expressive editorial C/S-spine mascot proportions. Best for articles, blogs, and clean technical explainers.
  • Mode 2.0 — Warm Watercolor Scene / Mixed-Media Studio (aspect_ratio="16:9"): Choose either (2a) cold-press #FAF8F3 watercolor illustration with wet-edge washes ({{ACCENT_HEX}} reserved for payoff) for storyboard concept plates, or (2b) mixed-media #FFFFFF tabletop with 1 real photographic physical object + flat 2D hand-drawn mascot interacting with it.
  • Mode 3.0 — Chibi Watercolor Dual-Cover Poster (3:4 vertical + 4:3 horizontal, or 16:9): Clean #FAF8F3 cold-press watercolor paper (or #FFFFFF high-contrast variant), bold brush-pen 2-line Chinese cover headline at top with a {{ACCENT_HEX}} watercolor highlight bar + optional compact sub-line, hero 2.5-head Chibi mascot beside 1–2 iconic props, and a neat bottom-right calligraphic {{SIGNATURE_TEXT}} signature. Generate 3:4 first, then pass its reference ID in image_references to generate the matched 4:3 horizontal cover.

2. Lock IP Consistency and Generate

Embed the full {{MASCOT_VISUAL_LOCK}} block from references/ip-and-style-dna.md (populated from ip-profile.yaml) into the image generation prompt. For multi-image batches or paired 3:4 + 4:3 covers, always pass the first confirmed output in image_references so anatomy, eyes, and line weight stay identical.

3. Visual QA and Surgical Clearance Edit

Inspect every generated graphic against the Mode-specific QA checklist in references/prompt-templates-and-qa.md:

  • Verify clean background (zero unwanted outer border frame) and 100% accurate, legible Chinese/English strokes.
  • Cover Head/Ear-Clearance Check: On 3:4 and 4:3 covers, verify there is a generous clean gap between any top subtitle text and the top of the mascot's ears/head. If text touches the silhouette, run 1 surgical edit scaling the mascot/prop group to ~90% and shifting it slightly downward.

Gotchas

  • Never mix Mode 1.0 and Mode 3.0 style prompts: Mode 1.0 forbids Chibi proportions and paper texture to stay ultra-crisp for editorial articles; Mode 3.0 explicitly embraces 2.5-head Chibi proportions and #FAF8F3 cold-press watercolor texture for high-CTR social covers.
  • Never render realistic SaaS UI screenshots: Draw Cursor, Claude Code, ChatGPT Canvas, DeepSeek-R1, MCP, 豆包, Muse, Grok Bot, and Agent CLI as simplified hand-drawn pen UI wireframes matching each product's recognizable layout signature.
  • Chinese text overload causes typos: Keep cover headlines to 2 short punchy lines (4–7 chars each) plus at most 1 compact sub-line, and body labels to 2–5 common characters.

Related Skills