paper-2-web — Claude Code skill: paper to site, poster and video
Turns an academic paper in LaTeX or PDF into an interactive homepage, a conference poster and a presentation video. Handy before a talk or when promoting a preprint.
npx claude-code-templates@latest --skill scientific/paper-2-webWhat paper-2-web does
The skill drives the Paper2All pipeline: it pulls content out of the paper with an LLM, generates layouts and refines them over several passes. Three outputs come out of it. Paper2Web builds a responsive multi-section page with interactive figures and tables. Paper2Video assembles slides, synthesises narration and syncs cursor movement; a talking head can be added through Hallo2. Paper2Poster produces a poster at any size, with templates, institution branding and a QR code for links.
First run needs Paper2All itself: clone the repo, create a conda environment on Python 3.11 and install requirements.txt. Put OPENAI_API_KEY in .env; logo search also wants GOOGLE_API_KEY and GOOGLE_CSE_ID. On the system side you need LibreOffice for document conversion and Poppler utilities for PDF handling.
Talking-head video needs an NVIDIA GPU with 48GB, so that mode is off the table on a laptop. The main pipeline runs through pipeline_all.py, the lighter video path through pipeline_light.py. The skill can call scientific-schematics to draw diagrams the paper is missing.
What it helps with
- build an interactive homepage for a preprint
- produce a conference poster at a custom size
- record a video abstract with narration and slides
- generate a full set of talk materials in one run
- add schematics a paper does not have
Example prompts for the agent
- Turn this paper into an interactive site with section navigation
- Generate a 60 by 40 inch poster from my LaTeX file
- Build a narrated video from this paper
How to install paper-2-web
1claude-code-templates
npx claude-code-templates@latest --skill scientific/paper-2-web2Skills CLI (Vercel)
npx skills add davila7/claude-code-templates --skill paper-2-web -a claude-code3Manually: put the files in place
~/.claude/skills/paper-2-web/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/paper-2-web/
Manually: put the files in place
.agents/skills/paper-2-web/SKILL.mdFor all projects use ~/.agents/skills/
Manually: put the files in place
.cursor/skills/paper-2-web/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install davila7/claude-code-templates paper-2-web2Manually: put the files in place
.github/skills/paper-2-web/SKILL.mdPlatform CLI
gemini skills install https://github.com/davila7/claude-code-templates --path cli-tool/components/skills/scientific/paper-2-webManually: put the files in place
.devin/skills/paper-2-web/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about paper-2-web
How do I install paper-2-web in Claude?
Through the marketplace: run npx claude-code-templates@latest --skill scientific/paper-2-web, then npx skills add davila7/claude-code-templates --skill paper-2-web -a claude-code. The skill file lands at ~/.claude/skills/paper-2-web/SKILL.md.
What does paper-2-web do?
It converts a paper in LaTeX or PDF into three formats: an interactive website, a presentation video and a conference poster. Under the hood it extracts content with an LLM and refines the design over iterations.
What do I need to run paper-2-web?
A conda environment on Python 3.11 with the Paper2All dependencies, OPENAI_API_KEY in .env, LibreOffice and Poppler utilities. Talking-head video requires an NVIDIA GPU with 48GB.