claudeers.
// Claude Plugins

paper2video

Claude Code skill: arXiv paper to a narrated, fact-checked explainer video (Three.js scenes, local Kokoro TTS, ffmpeg).

// Claude Plugins[ cli ][ api ][ claude ]#claude#plugins◷ MIT$open-sourceupdated 23 days ago
Actively maintained
100/100
last commit 11 days ago
last release 12 days ago
releases 1
open issues 0

Install with your AI

Paste into Claude Code, Cursor, or any agent — it reads the repo and wires the tool into your project.

Install and set up paper2video (claude-plugin project) into my current project.
Found on https://claudeers.com/paper2video
Repo: https://github.com/SunandBean/paper2video
Homepage/docs: —
Detected install method: claude-plugin → /plugin install paper2video@SunandBean/paper2video
Category: plugins. Platforms: cli, api.
Read the repo's README for exact setup and env vars, then install it and wire it into my project.

Claudeers Health Verdict:
active; community-verified: false. Confirm the source before running anything.
// or install directly (claude-plugin)
/plugin marketplace add SunandBean/paper2video
/plugin install paper2video@SunandBean/paper2video
// or clone
git clone https://github.com/SunandBean/paper2video

// compatibility

Platformscli, api
Operating systems—
AI compatibilityclaude
LicenseMIT
Pricingopen-source
LanguageJavaScript

Get your FREE $2.50 API credits to access TickAtlas financial data ↗

paper2video

A Claude Code plugin (one skill, /paper2video): arXiv paper in, narrated and fact-checked explainer video out.

/paper2video https://arxiv.org/abs/2610.30340

paper2video demo: depth to 3D points, query-based detection, BEV occupancy

Clips from the bundled example: a 4.5-minute explainer of NVIDIA's Foundation4D (CC BY 4.0), made end to end by this skill. All visuals are generated; every on-screen number traces back to a quote or table row in the paper.

All 14 scenes (contact sheet)

Foundation4D explainer contact sheet

  • Scenes: Three.js (3D) + canvas (2D), timed to the narration; reusable templates plus custom scenes per paper
  • Narration: English by default (Kokoro, local, Apache-2.0); Korean and 30 other languages with ko, ja, ... (Supertonic 3, local, OpenRAIL-M). No API keys
  • Cross-check: every number and claim is tied to a quote or table row in the paper and checked mechanically against the paper text and against every string the renderer actually drew (scripts/check_claims.py), then by an independent verifier subagent (references/verification.md)
  • Output: 1080p30 MP4, contact sheet, verification report. It never uploads or posts anything.

Install

Tested so far with Claude Opus 5.5 at medium effort; other models and effort levels have not been tried yet.

Requires Claude Code, git, Node 18+, Python 3.10-3.13 and about 1 GB of disk (Option B also uses curl). macOS or Linux; Korean videos on Linux need Noto Sans CJK.

Option A: plugin (recommended). In any Claude Code session:

/plugin marketplace add SunandBean/paper2video
/plugin install paper2video@sunandbean

That's it. The first /paper2video run installs the tools (Node packages, ffmpeg, headless Chromium, Python venv, voice model; a few minutes, once). Updates: /plugin marketplace update sunandbean, or turn on auto-update for the marketplace in /plugin → Marketplaces.

Option B: one-line script. In a terminal:

curl -fsSL https://raw.githubusercontent.com/SunandBean/paper2video/main/install.sh | bash

This clones into ~/.claude/skills/paper2video and installs the tools right away. Run the same line again to update.

Either way, your jobs and finished videos live in ~/paper2video/ and survive updates. The tools live separately (plugin data folder for A, .tools/ inside the clone for B), so updating never re-downloads the 1 GB toolchain.

Use

In any Claude Code session:

/paper2video <arxiv link or id> [overrides]
/paper2video 2610.35607 3 min, voice am_michael
/paper2video 2610.35607 한국어

The skill only runs when you call it (disable-model-invocation: true); pasting a paper link in a normal chat does nothing special.

Finished videos appear in ~/paper2video/videos/ with readable names (e.g. 2610.30340_Foundation4D.mp4); working files stay in ~/paper2video/jobs/<id>/. Override with P2V_HOME (both), P2V_OUT (videos) or P2V_JOBS (jobs).

Everything the skill does is also a terminal command, p2v in the skill folder:

p2v doctor                 # folders and what is installed
p2v smoke [--ko]           # toolchain test, should end with ALL PASS
p2v test                   # fast regression tests for the checker and the local server
p2v run <arxiv link>       # unattended run through `claude -p` (for cron or another bot)
p2v help                   # fetch, check, tts, preview, render, collect, setup

What's inside

  • SKILL.md workflow and defaults the agent follows
  • engine/ timeline player, 3D/2D primitives, templates
  • p2v single command for every step; scripts/ fetch (with license detection), TTS, claims check, render/mux
  • .claude-plugin/ plugin and marketplace manifests; install.sh for the git-clone route
  • references/ verification layers, scene authoring, licensing
  • local/ optional, git-ignored: put your own extra steps here (e.g. a publishing checklist); the skill runs them after its final report
  • jobs/2610.30340/ worked example (Foundation4D, NVIDIA, CC BY 4.0), jobs/_smoke/, jobs/_smoke_ko/ toolchain tests

License: MIT (see LICENSE). The example job includes text from the Foundation4D paper (arXiv 2610.30340, NVIDIA, CC BY 4.0).

Third-party: three.js (MIT), ffmpeg-static (downloads a GPL-3.0 FFmpeg build at install), kokoro-onnx (MIT) and Kokoro-82M weights (Apache-2.0), supertonic (MIT code, Supertonic 3 weights under OpenRAIL-M), Playwright (Apache-2.0). Korean subtitles need a Korean font (built into macOS; on Linux install Noto Sans CJK).

// faq

What is paper2video?

Claude Code skill: arXiv paper to a narrated, fact-checked explainer video (Three.js scenes, local Kokoro TTS, ffmpeg).. It is open-source on GitHub.

Is paper2video free to use?

paper2video is open-source under the MIT license, so it is free to use.

What category does paper2video belong to?

paper2video is listed under plugins in the Claudeers registry of Claude-compatible tools.

3 views
★ 49 stars
unclaimed
updated 23 days ago

// embed badge

paper2video on Claudeers
[![Claudeers](https://claudeers.com/api/badge/paper2video.svg)](https://claudeers.com/paper2video)

// retro hit counter

paper2video hit counter
[![Hits](https://claudeers.com/api/counter/paper2video.svg)](https://claudeers.com/paper2video)

// reviews

// guestbook

0/500

// related in Claude Plugins

🔓

A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.

// pluginsmultica-ai/★ 230,631[ claude ]
🔓

Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explainin…

// pluginsanthropics/⟨Python⟩★ 156,472[ claude ]
🔓

"CLI-Anything: Making ALL Software Agent-Native" -- CLI-Hub: https://clianything.cc/

// pluginsHKUDS/⟨Python⟩★ 58,467◷ Apache-2.0[ claude ]
🔓

financial-services — a Claude ecosystem project on GitHub.

// pluginsanthropics/⟨Python⟩★ 41,409◷ Apache-2.0[ claude ]
→ see how paper2video connects across the ecosystem