Exécutez n'importe quel Skill dans Manus
en un clic

Exécutez n'importe quel Skill dans Manus en un clic

html-to-svg

Étoiles12

Forks1

Mis à jour23 juin 2026 à 03:06

Turn a report into a portable SVG by authoring a structured/graphical HTML layout and rendering it to SVG with satoru-render. Use when the user asks to compile, summarize, or visualize a report, or wants graphical output that stays viewable in environments that cannot render HTML (chat, PDF viewers, image-only tools, git diffs). Proactively offer SVG output whenever you produce a report that benefits from layout, charts, tables, or styling.

Installation

Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.

Exécuter dans Manus

Source

ushironoko

ushironoko/dotfiles

Ouvrir le dépôt GitHub Voir les dépôts du créateur

Téléchargement

Exécuter dans Manus

Métiers associésSOC

Basé sur la classification professionnelle SOC

Développeurs de logicielsProfessions informatiques et mathématiques·SOC 15-1252

Explorateur de fichiers

4 fichiers

SKILL.md

readonly

name

html-to-svg

description

html-to-svg

Author a report as HTML (structure + graphics, your judgment), then render it to a single self-contained SVG with satoru-render. The SVG embeds every glyph as a vector path, so the result is portable: it renders identically anywhere an SVG can be shown, with no font or HTML engine required to view it.

When to use

The user asks to "compile / summarize / write up a report" and the content benefits from layout: cards, KPIs, tables, bar/progress charts, timelines, sections.
The user wants a graphic that survives environments without an HTML renderer.
Proactively: when you are about to output a structured report, offer to also produce an SVG version. Confirm before doing extra work if the user only asked for text.

Do not use it for plain prose, code, or anything that reads fine as Markdown.

Prerequisite

satoru-render (a WASM HTML→SVG/PNG/PDF engine, no browser/Puppeteer) must be on PATH. Install once, pinned:

bun install -g satoru-render@1.0.13

Check with which satoru-render; install it if missing. It is npm-distributed, so per project convention it is installed with bun install -g, not mise/ubi.

Workflow

Author the HTML. Design a structured, graphical layout for the report content. Use inline <style> and CSS-only graphics (flexbox/grid, bars, tables). Follow the design format in references/design-format.md — structure-first and low-chroma: structure, hierarchy, and reading flow carry the meaning, not color (no saturated primaries; no color-only labels). references/html-authoring.md has a render-verified template that already implements it — start from that template.
- Omit <!DOCTYPE html> — start at <html> (see the gotcha below).
- Set font-family: 'IBM Plex Sans JP', sans-serif; (the standard font for this format; covers Latin + Japanese) — satoru fetches it from Google Fonts at render time. Write the HTML to a temp file, e.g. $TMPDIR/report.html.
Render to SVG. Output extension .svg selects the format automatically:
```
satoru-render "$TMPDIR/report.html" -o report.svg -w 760 --no-jsdom
```
- -w sets width; omit -h so height auto-fits the content.
- --no-jsdom is correct for static HTML (faster, no extra dependency).
Verify. Confirm the SVG was written and is non-trivial. If the report contains non-ASCII text, check the glyphs actually rendered — see the network caveat below.
Deliver. Surface the file with SendUserFile. Offer to clean up the temp HTML.

Critical caveat: fonts are fetched over the network at render time

satoru resolves non-system fonts (including the format's IBM Plex Sans JP) from Google Fonts during conversion. Claude Code's default Bash sandbox blocks those hosts, so Japanese and other web-font text silently disappears from the output (structure renders, text is blank) while the command still exits 0.

When the report has non-ASCII or web-font text, run the conversion with network egress to fonts.googleapis.com and fonts.gstatic.com (and cdn.jsdelivr.net if you use emoji) — i.e. run the satoru-render Bash call with the sandbox disabled. If glyphs are missing, this is almost always the cause. Details and the local-font workaround are in references/satoru-render.md.

References

references/design-format.md — the default design format: the seven rules (whitespace, structure, plain language, no color-coded labels, low-chroma palette, paragraph writing in ordered sections, gloss every symbolic label), the palette ramps, typography scale, and the patterns that follow them.
references/satoru-render.md — full CLI/API reference, all flags, the font/network requirement, the DOCTYPE gotcha, and other output formats.
references/html-authoring.md — a render-verified report HTML template (implementing the design format) and authoring guidance (CSS-only charts, Japanese fonts, layout patterns).

Plus depuis ce dépôt

même dépôt

start-work

ushironoko/dotfiles

Start implementation work with worktree isolation and cross-session file conflict avoidance. Use this skill when beginning any non-trivial code change: after plan mode, when creating a new branch, implementing features, fixing bugs across multiple files, or refactoring. Also use when the user mentions worktree, bit issue, session coordination, or parallel work.

2026-07-1012

plan-review

ushironoko/dotfiles

Plan modeで作成したプランに対して、プロジェクト特性を分析し適切なレビューエージェントを自動選択・並列起動するスキル。引数なしで実行可能。

2026-06-0412

write-session

ushironoko/dotfiles

Persist mid-session progress (plan refinements, task progress notes, scope changes, decisions, blockers) to the bit issues created by start-work. Use as a checkpoint before context compaction, after meaningful progress (multiple tasks done, plan refined, scope shifted), before pausing work, or when the user mentions 'save session', 'snapshot', 'checkpoint', 'update bit issue', or 'persist progress'.

2026-05-0812

restoring-session

ushironoko/dotfiles

Restore session state (plan, target files, in-flight tasks) from bit issues created by the start-work skill. Use when resuming work after Claude Code restart, when the in-memory task list is empty but bit issues exist for the current branch, or when the user mentions "restore session", "repair session", "pick up where I left off", or references a prior bit-issue-tracked session.

2026-05-0812

empirical-prompt-tuning

ushironoko/dotfiles

agent 向けテキスト指示（skill / slash command / task プロンプト / CLAUDE.md 節 / コード生成プロンプト）を、バイアスを排した実行者に動かしてもらい、両面（実行者の自己申告 + 指示側メトリクス）で評価して反復改善する手法。改善が頭打ちになるまで回す。プロンプトや skill を新規作成・大幅改訂した直後、またはエージェントの挙動が期待通りにならない原因を指示側の曖昧さに求めたいときに使う。

2026-04-2212

smart-compact

ushironoko/dotfiles

セッションのコンテキストを分析し、重要な情報を保持するためのプロンプトを生成して /compact コマンドの実行を支援するスキル。コンテキストウィンドウが逼迫してきた時や、セッションを整理したい時に使用。

2026-03-2412

name

html-to-svg

description

html-to-svg

When to use

The user asks to "compile / summarize / write up a report" and the content benefits from layout: cards, KPIs, tables, bar/progress charts, timelines, sections.
The user wants a graphic that survives environments without an HTML renderer.
Proactively: when you are about to output a structured report, offer to also produce an SVG version. Confirm before doing extra work if the user only asked for text.

Do not use it for plain prose, code, or anything that reads fine as Markdown.

Prerequisite

satoru-render (a WASM HTML→SVG/PNG/PDF engine, no browser/Puppeteer) must be on PATH. Install once, pinned:

bun install -g satoru-render@1.0.13

Check with which satoru-render; install it if missing. It is npm-distributed, so per project convention it is installed with bun install -g, not mise/ubi.

Workflow

Author the HTML. Design a structured, graphical layout for the report content. Use inline <style> and CSS-only graphics (flexbox/grid, bars, tables). Follow the design format in references/design-format.md — structure-first and low-chroma: structure, hierarchy, and reading flow carry the meaning, not color (no saturated primaries; no color-only labels). references/html-authoring.md has a render-verified template that already implements it — start from that template.
- Omit <!DOCTYPE html> — start at <html> (see the gotcha below).
- Set font-family: 'IBM Plex Sans JP', sans-serif; (the standard font for this format; covers Latin + Japanese) — satoru fetches it from Google Fonts at render time. Write the HTML to a temp file, e.g. $TMPDIR/report.html.
Render to SVG. Output extension .svg selects the format automatically:
```
satoru-render "$TMPDIR/report.html" -o report.svg -w 760 --no-jsdom
```
- -w sets width; omit -h so height auto-fits the content.
- --no-jsdom is correct for static HTML (faster, no extra dependency).
Verify. Confirm the SVG was written and is non-trivial. If the report contains non-ASCII text, check the glyphs actually rendered — see the network caveat below.
Deliver. Surface the file with SendUserFile. Offer to clean up the temp HTML.

Critical caveat: fonts are fetched over the network at render time

References

references/design-format.md — the default design format: the seven rules (whitespace, structure, plain language, no color-coded labels, low-chroma palette, paragraph writing in ordered sections, gloss every symbolic label), the palette ramps, typography scale, and the patterns that follow them.
references/satoru-render.md — full CLI/API reference, all flags, the font/network requirement, the DOCTYPE gotcha, and other output formats.
references/html-authoring.md — a render-verified report HTML template (implementing the design format) and authoring guidance (CSS-only charts, Japanese fonts, layout patterns).