<!-- xid: 91B67D4AC2F3 --> <a id="xid-91B67D4AC2F3"></a>
Scanned 9/12/2026
Install to Claude Code
npx -y skills add synthaicode/XRefKit --skill marketing_slide_png --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Marketing Slide Png?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/synthaicode-marketing-slide-png)More formats (shields.io, HTML) on the badges page.
<!-- xid: 91B67D4AC2F3 -->
<a id="xid-91B67D4AC2F3"></a>
# Skill: marketing_slide_png
## Purpose
Create slide visuals as CSS/HTML-rendered PNG assets for Marketing Group work. Use this when the deck should be figure-first: Markdown mainly controls page breaks and image placement, while slide titles, key messages, labels, and diagram structure live inside the generated PNG.
This skill may also be used in `single_image_infographic` mode when Marketing Group needs a standalone one-page repository explainer image, such as a current-state visual summary of XRefKit's operating model.
## References
- [Marketing Slide PNG Style Guide](references/style-guide.md#xid-8B1A7D45C9E2)
- [Japanese Presentation Tone Guide](references/japanese-presentation-tone.md#xid-D5C8A5F2E731)
- [Marketing Slide PNG Layout Patterns](references/layout-patterns.md#xid-B0F915C6A4D8)
- [Marketing Slide PNG Pattern Samples](references/pattern-samples.md#xid-3F2B8D7EA551)
- [XRefKit Slide Authoring Checklist](references/xrefkit-slide-authoring-checklist.md#xid-6A2D9F4C8E71)
## Ownership
- Treat this skill as Marketing Group work.
- Use PR Group for official announcement work instead of this skill.
## Inputs
- deck markdown path
- target asset directory
- slide-by-slide message list
- visual direction or brand constraints
- whether the deck should be image-based for all slides or only selected slides
- output mode:
- `slide_deck`
- `single_image_infographic`
## Outputs
- `diagram.css`
- `render.mjs`
- per-slide `*.html`
- per-slide `*.png`
- updated markdown deck that embeds the generated PNG files
- for `single_image_infographic`: one standalone PNG plus its reusable HTML/CSS/render sources
## Startup
- Confirm the target deck path.
- For `single_image_infographic`, confirm the target repository scope and target image path instead of a deck path.
- Confirm that the deck should be image-based, or identify which slides should use image-based composition.
- Confirm the audience and purpose so each slide can be reduced to one visual message.
- Confirm whether titles and supporting labels should live inside the PNG. Default to yes for image-based decks.
- Read the style guide and pick layout patterns before writing any render script.
- When the deck is Japanese, read the Japanese presentation tone guide and keep claims clear but wording natural.
## Planning
- Map each slide to one core message.
- For `single_image_infographic`, map the image to one central claim and a small number of visual sections.
- When the image represents the current repository, build a repository fact map before visual layout:
- entry points read
- repository structures shown
- control rules shown
- evidence-backed claims
- marketing explanations
- unknown or unstable claims
- Decide what stays in Markdown:
- page breaks
- image embed lines
- optional speaker notes only
- Decide what moves into the PNG:
- slide title
- subtitle or one-line claim
- boxes, arrows, comparisons, loops, labels, badges
- Prefer comparison, flow, hub-and-spoke, and loop layouts for concept slides.
- Choose patterns from `references/layout-patterns.md` instead of inventing a new structure by default.
- Keep one visual per slide. If the slide needs two different stories, split it.
- Keep one central claim per standalone infographic. If the image needs multiple unrelated claims, split it into a deck or a series of images.
## Execution
- Create or reuse an asset folder under `human-docs/ja/assets/<deck-name>/` when the deck is Japanese, or the matching human-docs tree when not.
- Create:
- `diagram.css` for shared visual rules
- `render.mjs` for deterministic HTML generation
- one `*.html` per slide image
- Put the slide title and key message inside the generated image when the deck is image-based.
- Keep the markdown deck minimal. Prefer this pattern:
```md
---

```
- Generate HTML first, then render PNG from those HTML files.
- Prefer reproducible local commands. In this repository, the working pattern is:
```powershell
node human-docs/ja/assets/<deck-name>/render.mjs
npx --yes playwright install chromium
npx --yes playwright screenshot --browser chromium --viewport-size "1600,900" file:///.../01_example.html .../01_example.png
```
- If multiple images are needed, render all HTML files in a loop and keep filenames ordered such as `01_`, `02_`, `03_`.
- For `single_image_infographic`, use an explicit filename that names the repository snapshot, such as `xrefkit_os.html` and `xrefkit_os.png`.
## Monitoring and Control
- Verify every slide still has one visual message.
- Verify every standalone infographic still has one central claim.
- For repository snapshot infographics, verify the image represents the current repository state and does not invent undocumented capabilities.
- Verify important XRefKit control ideas are not accidentally dropped when they are part of the requested message:
- AI does not rely on memory
- required context is loaded from control assets
- execution and checking are separated
- unsupported guessing is stopped
- unknowns and risks stay explicit
- context-direction guard protects higher-layer intent and authority
- trade-offs and approvals return to humans
- audit trail and XID traceability support reproducibility
- Verify markdown does not repeat text that already appears inside the PNG.
- Regenerate PNG files after any structural change to the diagrams.
- Keep CSS simple and consistent across slides:
- stable typography
- limited color palette
- consistent spacing
- minimal decorative effects
- Follow `references/style-guide.md` for color, typography, and markdown split decisions.
- Follow `references/japanese-presentation-tone.md` when writing Japanese titles, subtitles, badges, and summary lines.
- Apply `references/xrefkit-slide-authoring-checklist.md` before closure when the deck explains XRefKit or is published in the XRefKit site.
## Closure
- Confirm every embedded PNG exists on disk.
- For `single_image_infographic`, confirm the standalone PNG exists on disk.
- Confirm the markdown paths match the generated asset paths.
- Keep rerenderable sources:
- do not keep only PNG
- preserve CSS and HTML generation sources
- Preserve the rendering command path so another agent can rerun the asset generation without reverse-engineering the process.
## Rules
- Markdown should mainly own page breaks and image placement for image-based decks.
- For standalone repository infographics, preserve a short source note or session log that records which repository entry points were read.
- Put the communicative content into the image when using the image-based pattern.
- Do not duplicate the same title and bullets in both Markdown and PNG.
- Do not create over-decorated diagrams. Favor clarity over novelty.
- Do not use this skill for official announcements; that belongs to PR Group.
## Reporting Contract (共通報告)
- reporting_profile: summary_first
Use the shared [Skill Reporting Contract](../../docs/core/contracts/081_skill_reporting_contract.md#xid-6B2D9F4A1C73) in the final report. Start with these headings in this order:
1. Status — done, partial, blocked, or escalated
2. Result — what was produced or decided
3. Evidence — output, evidence, checks, or XIDs
4. Open Items — unresolved unknowns, risks, judgments, or なし
5. Handoff — next owner and next action, or なし
Keep this summary-first section visible before Skill-specific detail; do not omit empty sections.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!