Index of every page on this site: llms.txt
All skills
aktsmm avatar

/visualize-as-infographic

@efb4da6
by yamapanaktsmm/agent-skills27 stars
5
Run it yourself

Create colorful infographic PNGs from a conversation, topic, file, skill, or workflow. Use when: インフォグラフィック, 図解, 挿絵, 章扉, X投稿用画像, OGP, HTML→PNG, visualize session, poster image. Produces 2-3 self-contained HTML patterns, renders static PNGs, and verifies every image visually before reporting. Not for animated GIFs, architecture/sequence/flow diagrams (editable .drawio or interactive HTML), or HTML slide decks.

Recorded with Codex, gpt-6.1-sol, medium effort

Prompt

Create two colourful 1200x675 infographic patterns explaining a community tool library's four lending rules: reserve one tool, collect with ID, return within seven days, report damage. Use English, dark backgrounds, no signature, no numeric impact claims, and distinct visual directions. Follow the Skill: independent HTML files, render both PNGs, inspect both and fix defects. Also create index.html that embeds both final posters as inline images in a responsive gallery, with a working pattern selector and concise descriptions. Don't ask questions. Make the design calls yourself. Use TypeScript for build tooling if needed. Save the final demo as one self-contained index.html with all CSS and runtime JavaScript inline. Bundle any package code locally. All visible controls must work. It will run in an opaque iframe with sandbox allow-scripts only: no form submissions, storage, popups, or downloads. Use explicit button handlers and in-page results. Support reduced motion, keyboard use, 390px mobile, and 1440px desktop. Verify the important controls in a browser before finishing.

Desktop screenshot of the page the Agent built for: Create two colourful 1200x675 infographic patterns explaining a community tool library's four lending rules: reserve one
Phone screenshot of the page the Agent built for: Create two colourful 1200x675 infographic patterns explaining a community tool library's four lending rules: reserve one

Recorded on an older version of this Skill.

Use this Skill: https://skilld.dev/gh/aktsmm/agent-skills/visualize-as-infographic

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈121 tokens for metadata: the name and description. ≈1.8k when used: this file.

Description uses 5.5% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈439 characters, or 5.5%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Visualize As Infographic

Turn a conversation summary, topic, file, skill, or workflow into 2-3 polished infographic PNGs. Generate one self-contained HTML file per pattern, render PNGs, inspect every PNG, fix visual defects, and report usable outputs.

When To Use

  • User asks for インフォグラフィック, 図解, 挿絵, 章扉, X投稿用画像, OGP, visualize, or HTML→PNG.
  • User wants multiple visual directions, not a single text-only slide.
  • The artifact should be shareable as PNG and reproducible from HTML.

Inputs

  • Target: explicit argument if provided; otherwise summarize the latest relevant session context.
  • Usage: default is X / SNS. If the user mentions book chapter inserts, OGP, slide, or PDF, adapt the ratio and style.
  • If brightness, signature/title text, or aspect ratio are unspecified, ask once before authoring with compact choices:
    • Brightness: default dark / brighter / white background
    • Signature or title mark: default none / small mark / specify text
    • Ratio: default X multi-pattern / 16:9 only / square only
  • Treat default as the normal fallback when the user says default or gives no preference. Do not use a separate omakase option when it means the same thing as default.
  • Ask additional clarifying questions only if the output would materially change.

Output Defaults

  • Create 2-3 different patterns, one HTML per pattern.
  • Save under output/<topic>-diagrams/.
  • File names:
    • HTML: <topic>-<pattern>.html
    • PNG: <topic>-<pattern>.png
  • Common ratios:
    • X / SNS: 1200x675 and/or 1200x1200
    • Blog OGP: 1200x630
    • Chapter insert / book figure: match the book tone; prefer calm, high-readable layouts over flashy decoration.

Design Rules

  • Use visual hierarchy: title, short lead, 3-6 visual units, concise footer if useful.
  • Use a consistent licensed icon set for generic concepts and embed its SVG/image assets rather than loading a runtime CDN. For product identities, use permitted official assets, not generated logos.
  • Match explanatory labels to the artifact's audience language. For Japanese articles, use natural Japanese for titles, captions, status labels, and generic UI; keep product names, API fields, and other proper nouns in their official form. Avoid literal labels that sound unnatural, such as 段階で採点, when 定義した基準のどの段階に近いか is the intended meaning.
  • Read numeric claims from the source artifact or SSOT instead of retyping them. For latency, cost, accuracy, and token values, state the measurement boundary in or next to the image: per request, serial total, fresh-process end to end, estimate, or saved-result playback.
  • If an infographic depicts motion or elapsed time, state inside the asset whether movement is illustrative or timed from measurements. Animation duration must not be presented as API or processing latency unless frames are generated from those timestamps.
  • Avoid information overload. If text looks dense, split into another pattern rather than shrinking aggressively.
  • Set html, body to exact poster width / height and overflow: hidden.
  • Use Noto Sans JP for Japanese poster text unless a project-specific style says otherwise.
  • Use stable dimensions for cards, badges, rails, and panels so text or icons cannot push adjacent elements.
  • For chapter inserts, match the book's palette and mood; do not force SNS-dark neon styling if the publication design is quiet.
  • Do not add signatures, account names, license text, or skill names unless the user asks.
  • Do not render usage labels or production notes such as X main, best use, or list + highlight in image corners. Keep usage and recommendation notes in the completion report only.

Rendering Procedure

  1. Create the output folder.
  2. Create independent HTML files. Do not pack multiple patterns into one HTML.
  3. Render with Playwright. Prefer Python Playwright if available. A helper script is provided: render_infographics.py. If Playwright MCP blocks file://, serve the output folder with python -m http.server <port>.
  4. Use viewport-size clipping, not fullPage, when decorative elements can overflow the poster. fullPage may capture off-canvas decoration and create white margins.
  5. Use domcontentloaded plus a short fixed wait if external fonts make load or networkidle hang.
  6. Use device_scale_factor=1 when the PNG must be the stated size. Use 2 only when high-density output is desired and the larger pixel dimensions are acceptable.
  7. Prefer container-type: inline-size when only width-relative units such as cqw are needed. With container-type: size, give the wrapper a definite block size or an intrinsic fallback: full size containment ignores descendants when resolving an auto block size, so the wrapper can collapse to an empty-box height while a descendant's scrollHeight stays nonzero.
  8. Inspect every PNG with view_image.
  9. Inspect at the target embed width as well as full size. A 1200px image that is readable full-size can become illegible when embedded near 360px on mobile.
  10. If any PNG has overlap, clipping, unreadable text, wrong order, unexpected margins, wrong pixel size, or excessive density, fix the HTML or renderer, render again, and re-check that PNG.
  11. Do not send final until all PNGs have completed inspect -> fix if needed -> rerender -> re-inspect.
  12. Clean up after finishing: stop any HTTP server, move stray PNGs from the workspace root into the output folder, keep the source HTML, and delete temporary render scripts unless they are deliverables.

Visual QA Checklist

  • No text overlaps other text, icons, rails, cards, or borders.
  • No content is clipped at the image edge.
  • No unintended white margins or oversized canvas from screenshot capture.
  • Pixel dimensions match the promised output size unless high-density output was explicitly chosen.
  • Step numbers and visual order match DOM / reading order. CSS grid grid-column / grid-row fixed placement ignores DOM order and is a common cause of scrambled numbering.
  • In named comparison plots, label points near their marks or connect them visibly to names at the target embed width; do not make readers decode numbered points from a distant legend alone.
  • Card text fits without cramped line breaks.
  • Contrast is readable on the final PNG, not just in HTML.
  • Generic English UI does not remain in a Japanese artifact unless it is part of a quoted product surface.
  • Numeric values and caveats agree with their source artifact, and an image shared alone cannot confuse illustrative playback with measured latency.
  • A blank or short capture is treated as a defect to diagnose, not an empty design. Check the wrapper's computed height before blaming the content.
  • The final answer refers to the current infographic request, not an older topic from the conversation.

Completion Report

Provide a compact table:

PNG Size Pattern Best use

Then recommend one image to use first and explain why in one sentence.

Do Not

  • Do not leave a known-bad PNG as a recommended output.
  • Do not invent numeric claims, quotes, or external facts that are not in the source material.
  • Do not use hardcoded local absolute paths inside generated HTML.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No alerts11d3 checks · Risk SAFE
  • Gen Agent Trust Hub11d

    This skill converts conversations or topics into infographic PNGs by generating HTML and rendering it with a provided Python script using Playwright. It follows secure practices like local rendering and embedding assets.

  • Socket11d

    No alerts

  • Snyk11d

    Risk: LOW · No issues

Signed by skilld at efb4da6. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated last month
argument-hint
図示する対象、用途(X投稿/章扉/ブログOGP等)、比率やパターン数の希望
user-invocable
true
metadata
{
  "author": "yamapan (https://github.com/aktsmm)"
}

README badge

README badge for aktsmm/agent-skills/visualize-as-infographic