arxiv2zhtml
Translate an arXiv/official HTML paper into Simplified Chinese and output a fully offline single-file HTML.
Skill by Rvosuke
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/rvosuke-arxiv2zhtml
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: frontend, backend, research
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
- Runs as
- Python package
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/arxiv2zhtml
What's inside
- Skill name
arxiv2zhtml- Description
- Translate an arXiv/official HTML paper into Simplified Chinese and output a fully offline single-file HTML. Use when 翻译论文/arXiv HTML 为中文, inlining images and resources, or producing a self-contained Chinese HTML that preserves the original structure. Requires file access and Python 3; do not start an HTTP server.
- License
- MIT
- Instructions
- 34 lines in SKILL.md · view on GitHub
Before installing
- Ships executable scripts
Files (3)
- SKILL.md
- translation-guide.md · references/
- audit_html.py · scripts/
SKILL.md (first lines)
# Arxiv to Chinese HTML Translate a paper's official HTML into natural, faithful Simplified Chinese and output a single self-contained HTML file with no external dependencies, for offline reading. Structural and resource integrity comes before visual polish. ## Sources and boundaries - Prefer official HTML (`https://arxiv.org/html/<id>`; a bare id resolves to the latest version). Use official TeX to verify formulas or structure when the HTML is broken; PDF only for what neither source confirms. - Do not add facts, explanations, or citations absent from the paper. Sentences may be split, merged, or reordered, but facts, numbers, negations, comparisons, and claim strength must not drift. - Preserve the section tree, formulas/MathML, tables (cell order, `rowspan`/`colspan`, values), figures and captions, algorithms, footnotes, appendices, references, and link relations. Replace only text nodes; do not regenerate the page or change IDs. - Keep dataset, benchmark, task suite, model, robot/arm platform, simulation environment, repo/handle, metric label, and class label **names in English** — do not translate them mechanically. Keep field-convention English terms such as token and embedding as well. Do not translate formulas, code, URLs, or labels. - Never start an HTTP server or listen on a port; preview locally with `file://` only. ## Workflow 1. **Translate** section by section in natural academic Chinese word order (subject–method–condition–result), avoiding word-by-word mapping and English inversion. Use full-width punctuation in Chinese; keep formulas, code, and English abbreviations in their original form. Treat tables, captions, footnotes, and appendices as first-class content, not optional. Keep the proper nouns listed above in English as you encounter them; do not pre-scan the whole paper to build a glossary. 2. **Inline resources**: inline CSS into `<style>` and any necessary JS into `<script>` (prefer CSS over script); download paper images and encode them as `data:` URIs with the real MIME type; do not reference external fonts. Check every `src`/`srcset`/`href`/CSS `url()`/`@import` — rendering resources must not hit the network (`#fragment` and `data:` are fine). Remove site navigation, feedback, and telemetry controls, but not internal paper navigation, footnote backlinks, or reference links. 3. **Save** to a new file; do not overwrite the original HTML (a `_zh` or version suffix is enough — no hashes or baseline ceremony). 4. **Structure audit**: run the deterministic script to verify structure and resources are conserved: ```bash python3 scripts/audit_html.py original.html translated.html --json audit.json ``` It checks DOM/IDs, MathML hashes, table grids, image data URIs and decoded hashes, bibliography/footnote/citation edges, internal link targets, and external dependencies. If an unauthorized structural change appears, stop and fix it rather than widening an allowlist to hide the diff. In translation mode a large visible-text diff is expected; focus on the structural items. 5. **Spot-check**: open the file with Chromium/Chrome via `file://` at desktop width once: formulas/MathML render readably, images show offline, tables do not overflow the page, and there is no clipped or low-contrast text. If Chromium is unavailable, state that visual checking was not done rather than passing off static parsing as a visual pass. ## Optional (only when asked) - **Numbered citations**: when converting author-year citations to `[n]`, assign stable numbers by bibliography order; keep every number as an internal link to its entry. Runs may render as ranges but all member links must remain reachable. - **Proper-noun fix-up / layout-only mode**: for an existing Chinese draft, change only mistranslations confirmed against the original, no drive-by polishing. In layout-only mode freeze all visible text and touch only containers, CSS, and accessibility attributes. Translation details are in [references/translation-guide.md](references/translation-guide.md).
From Rvosuke/Skills, MIT.
More from Rvosuke/Skills
- 0.586score
writing
Rvosuke
撰写、改写和润色专业文稿,优化标题、叙述主线、语序与标点。适用于项目介绍、进展报告、研究说明、交接材料,以及评论整理和多页文档组织。
SkillAFreshness: A★ 1Claude Code +9 - 0.549score
arxiv-blog-skill
Rvosuke
Read an arXiv paper and write a Chinese popular-science blog that explains the topic and method in depth.
SkillBFreshness: B★ 1Claude Code +9 - 0.549score
ml-coding
Rvosuke
Write, debug, and review ML research code.
SkillBFreshness: B★ 1Claude Code +9 - 0.511score
handoff
Rvosuke
Compact the current conversation into a handoff document for another agent to pick up.
SkillBFreshness: B★ 1Claude Code +9 - 0.467score
grill-me
Rvosuke
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree.
SkillCFreshness: C★ 1Claude Code +9
Signals
- Stars
- 1
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Aug 27, 2026 (2 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/Rvosuke/Skills
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from Rvosuke
- 0.586score
writing
Rvosuke
撰写、改写和润色专业文稿,优化标题、叙述主线、语序与标点。适用于项目介绍、进展报告、研究说明、交接材料,以及评论整理和多页文档组织。
SkillAFreshness: A★ 1Claude Code +9 - 0.549score
arxiv-blog-skill
Rvosuke
Read an arXiv paper and write a Chinese popular-science blog that explains the topic and method in depth.
SkillBFreshness: B★ 1Claude Code +9 - 0.549score
ml-coding
Rvosuke
Write, debug, and review ML research code.
SkillBFreshness: B★ 1Claude Code +9 - 0.511score
handoff
Rvosuke
Compact the current conversation into a handoff document for another agent to pick up.
SkillBFreshness: B★ 1Claude Code +9 - 0.467score
grill-me
Rvosuke
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree.
SkillCFreshness: C★ 1Claude Code +9
Cite this page
arxiv2zhtml skill. Agent Mart. https://agentmart.dev/skills/rvosuke-arxiv2zhtml (last updated 2026-10-24)