Skills

papercut-explainer: batch-produce vertical papercut-style explainer videos in Claude Code

papercut-explainer works best as a staged pipeline in Claude Code, from strategy to QA, producing vertical papercut-style explainer short videos. The catch: you must have Node 20+, ffmpeg, rsync, and Python 3 set up, and full features require cloning the repo and running install.sh — not just installing the skills.

Official source · papercut-explainer

Install

npx skills add 0xS4N4TG/papercut-explainer

papercut-explainer is a set of Claude Code skills plus a Remotion library for producing vertical papercut-style explainer short videos. It is not a single prompt, but a pipeline from topic selection to finished video: strategy, hook, script, storyboard, voice, build, and QA — each stage handled by its own skill, plus a papercut-scene subagent. Every stage writes a file, so state lives outside the chat: you can stop midway, make changes, and pick up where you left off.

Who it's for

  • Creators who want to use Claude Code to batch-produce or continuously produce short vertical explainer videos, and who are willing to go through the full stage pipeline rather than expecting a one-shot video.
  • Prerequisites you need or must install: macOS (or Linux/WSL), Node 20+, npm, ffmpeg, rsync, and Python 3. The optional voice virtual environment requires Python 3.10–3.12; rendering SFX/music requires Chrome/Chromium.
  • Not suitable for: users who don't use Claude Code as the front end; those without the runtime environment above. The evidence contains no information about it working on native Windows (non-WSL), so that is unknown.

First attempt in 3 minutes

The official docs offer two installation paths; the recommended main path is the fastest skills-only install:

  1. Prerequisites: Claude Code is working, Node 20+ and npm are in place.
  2. In a terminal, run: npx skills add 0xS4N4TG/papercut-explainer. This installs only the skills.
  3. For a full install (including the Remotion library), the official method is to clone the repository and run install.sh. The installer symlinks the skills into ~/.claude/skills, symlinks the papercut-scene subagent into ~/.claude/agents, and writes ~/.config/papercut-explainer/config.env (containing PAPERCUT_LIB and the Python path it creates); it does not install system packages or modify your shell profile.
  4. Kick off a low-risk task in Claude Code: simply say "use papercut-explainer to do a strategy-stage topic analysis" — the router skill will hand the task to the strategy skill.
  5. Success signal: the file written by that stage appears on disk — the official docs explicitly state that every stage writes a file, so you can see the strategy-stage output outside the chat. The file location and exact filename are not stated in the evidence and remain unknown.

It's safest to stop at the strategy stage on your first attempt: it only produces analysis text and does not render video. The subsequent hook, script, and storyboard stages likewise produce files, while the build stage actually renders video (requires ffmpeg, etc.), and voice may create a Python virtual environment — confirm the side effects before running these.

Use cases

Case 1: Early-stage strategy for a product explainer. Start a task in Claude Code that runs papercut-explainer's strategy stage; the router will dispatch the task to the corresponding skill. The confirmable result is a strategy artifact written to a file; details such as the wording of hook copy or the exact fields of the strategy file are not given in the evidence, so rely on the actual output.

Case 2: An interruptible long pipeline using per-stage files. For the same video, advance hook→script→storyboard in sequence — each stage writes one file; if you want to change the tone midway, just edit the corresponding stage's file and let the agent continue to the next stage. The official docs explicitly state "Every stage writes a file, so you can stop, edit and resume", which is the direct basis for this resumable workflow. The voice (dubbing) and build (finished video) stages depend on ffmpeg, the Remotion library, etc., so they require a full install with dependencies in place before they can run.

Usage boundaries

  • File writes are by design: every stage writes a file, and the installer writes ~/.config/papercut-explainer/config.env and symlinks ~/.claude/skills and ~/.claude/agents. Whether the skills themselves read or write files beyond the pipeline, execute commands, or access the network is not stated in the evidence and remains unknown.
  • Hard runtime requirements: Node 20+, npm, ffmpeg, rsync, Python 3; the optional voice venv requires Python 3.10–3.12; SFX/music rendering requires Chrome/Chromium. The front end is Claude Code. The official requirement is macOS or Linux/WSL.
  • Officially stated limits: the full-install install.sh does not install system packages and does not modify your shell profile.
  • Not yet confirmed: supported languages (the evidence does not address differences between Chinese and English finished videos), the exact naming and locations of per-stage files, and which stages are available under the skills-only path without a full install. Verify these against the repository's latest README before actual use.