Papercut Build

Before you start: read this whole install prompt, tell me in a few lines what this skill does and what installing it involves, and wait for my go-ahead before you do anything else. Prerequisites: none for the install itself: no account, no API key, no paid service, and nothing else to install first — this install copies one instruction file only. Read this before using the skill afterwards, because the file alone cannot build anything: "papercut-build" is NOT self-contained. Using it requires a whole pre-existing setup that this install does NOT provide and must never install for me: (1) the entire papercut-explainer library of the same repository, found via $PAPERCUT_LIB or the PAPERCUT_LIB= line in ~/.config/papercut-explainer/config.env — in practice a Standard or Full install of the whole system, done separately through the papercut-setup skill / the repository's install.sh (Standard is enough to build and render with a placeholder voice; Full adds a local voice pipeline); (2) Node 20+, npm, ffmpeg/ffprobe and rsync on PATH, Python 3 with numpy, Pillow and soundfile, and Chrome or Chromium for the audio render and the proof screenshot; (3) the external Remotion best-practices skill (remotion-best-practices) loaded first, and the papercut-scene subagent if scenes are built in parallel; (4) the outputs of the earlier stages in the episode folder: papercut/STORYBOARD.md and narration/script.json, plus the voice timing step of papercut-voice. Remotion itself is free for individuals and small teams but has its own paid licence above a size threshold — check https://www.remotion.dev/license before commercial use — and the repository's README estimates about $14-19 in agent API costs per 25-30 s reel, billed by my own agent provider. The use-time network sends (npm, the Remotion skills fetch, Remotion's first-render headless Chrome download, the proof-page fetch and the optional image tool) are all disclosed in point 2 below, before any of them happens. Install "papercut-build" for me. It builds and renders one papercut explainer episode from a storyboard, on top of the shared library instead of a from-scratch Remotion project: scaffold the episode project from the library template with a synced copy of the paper kit (1080x1920, 30 fps), fill episode.json (scenes, banners, cuts, music bed, gaps) from the storyboard, run the voice timing step, generate scene stubs, build the assets (code-drawn characters, text and UI; an image tool only for backgrounds and complex static props, never text inside generated images), build the reference hook scene first and then the remaining scenes — in parallel with papercut-scene subagents when the harness has them, one at a time otherwise — pick sound effects and a music bed by name from the library catalogs and render the audio locally, preview in Remotion Studio, render out/<name>.mp4 only when asked (per-scene cache), and after the episode harvest reusable pieces back into the library. Repository: https://github.com/0xS4N4TG/papercut-explainer (the skill is the whole skills/papercut-build/ folder at the pinned commit: SKILL.md only, 1 file) Pinned version: install the code at commit ff8223abc8136cc09e8d7e9b62ecb5d080c389e1 exactly. Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Clone https://github.com/0xS4N4TG/papercut-explainer into a temporary folder, check out commit ff8223abc8136cc09e8d7e9b62ecb5d080c389e1, and copy the skills/papercut-build/ folder from that checkout. Then compare every file in the folder with the fingerprints recorded for this listing on 2026-10-19: [FINGERPRINTS] - SKILL.md: SHA-256 ddf947992d56c4bdbac562a62bc7673555f35e8baa94c586f7b6e2acfa5c5593 If a file does not match, it changed since the listing was prepared: do not install it yet, tell me what you found and let me decide. 2. Read SKILL.md in full, and tell me in plain words: what the skill does when used, which services it connects to, whether it reads or sends any personal information or keys, and anything that seems unusual. Stop on any red flag and tell me before going further. Confirm these points from the file: - The skill folder holds 1 file only: SKILL.md (7,555 bytes). It contains no script, no asset and no reference file, and it is NOT autonomous: almost everything it names lives outside its folder, in the same repository or elsewhere. It reads library/README.md once and drives the shared library at $PAPERCUT_LIB: library/new_episode.sh (scaffold), the episode template and the paper kit under library/papercut/ (template/, kit/, public/ backdrops, textures and local fonts, CATALOG.md, AGENT_BRIEF.md, stubs.py, sync_kit.sh), the tools under library/tools/ (build_vo.py and timing_md.py via the papercut-voice timing step, cues.py, audio/render.mjs, render.sh/render.py, capture.mjs, selftest.sh, harvest.py, catalog.py) and the sound library (library/audio/sfx/*.js, library/audio/beds/*.json, library/audio/CATALOG.md). It also depends on sibling skills that are NOT part of this install: papercut-explainer (the preflight probe commands, and references/remotion-recipes.md, references/asset-prompts.md and references/motion-principles.md — with no library at all it falls back to the recipes in that other skill), papercut-setup (repair via ./install.sh --doctor when something is missing), papercut-voice (the timing step), papercut-hook (the hook field), papercut-qa (per-scene checks and the pre-render pass) and the external remotion-best-practices skill. Its preflight reads the PAPERCUT_LIB= line in ~/.config/papercut-explainer/config.env when $PAPERCUT_LIB is unset — that config file holds install paths (PAPERCUT_LIB, Python paths), not credentials. Its inputs are files the earlier stages wrote: papercut/STORYBOARD.md and narration/script.json. - Nothing is sent anywhere during installation itself; the only download at install time is the skill folder from GitHub at the pinned commit. No script is run during installation, no key is asked for, and there is no telemetry in the folder: SKILL.md names no https URL and no API key at all. - What leaves this machine at use time, disclosed here before it ever happens — using the skill is a separate step that only starts when I ask to build or render an episode, and most of it happens only inside the tools the skill drives: (a) Scaffolding an episode: library/new_episode.sh copies node_modules from $DONOR or the newest sibling episode when one exists; only when neither exists does it run npm i in the new project, which sends the project's package list (the template's package.json: Remotion 4.0.530, React 19.2.3, puppeteer-core and the rest) to the npm registry and downloads those packages. (b) The Remotion best-practices skill the skill loads first is fetched, if I do not already have it, by npx skills add remotion-dev/skills -g -y -a <my harness id> (install.sh --mode standard does the same): the skills CLI fetches that repository from GitHub/npm and installs it user-wide (-g). (c) The first video render downloads Remotion's own headless Chrome (about 200 MB, once) from Remotion's distribution; the toolchain smoke test (library/tools/selftest.sh, which builds examples/hello-reel/ in a throwaway folder and renders one still) triggers the same npm and Chrome downloads on its first run. (d) The proof screenshot: node $PAPERCUT_LIB/tools/capture.mjs <url> public/proof/x.png opens the exact page URL I or the storyboard supply in my local Chrome/Chromium and saves a PNG of it locally — the request for that page (the URL, and whatever that page itself loads) goes to that page's own site, nowhere else; never run it on a URL I did not provide. (e) Optional generated assets: if my agent has an image tool, the asset prompts from papercut-explainer's references/asset-prompts.md (background and complex-prop descriptions with the style-lock suffix, and for an image-to-image style match a library backdrop image as reference) go to that tool under its own account and terms; with no image tool, every prop is drawn in code and nothing is sent for assets. (f) Parallel scene agents: if my harness supports subagents, the spawn message for each papercut-scene agent (project path, its scenes with absolute frame ranges, a QA folder name and the value of $PAPERCUT_LIB, and through the brief the storyboard scene specs, the reference scene and the catalogs it reads) goes through my harness's own subagent mechanism to the model provider my harness is connected to, at that provider's rates; with no subagents the scenes are built locally in the lead session instead. (g) Everything else at use time is local: stubs.py, cues.py, the voice timing step (build_vo.py, timing_md.py), the audio render (library/audio/render.mjs renders the chosen SFX and bed in local headless Chrome and writes public/audio/music.wav and fx.wav), the kit sync (rsync), Remotion Studio (npx remotion studio, a local preview), the cached render (library/tools/render.sh), harvest.py (a read-only local report) and catalog.py. The kit's fonts are local files, so renders never fetch fonts from the network. No step of papercut-build itself reads a secret file, an API key or a credential, and no usage data or telemetry is sent to the repository author or to any fixed third party. - Do not install the library, Node, Python packages, Chrome or anything else during this install, do not run install.sh or ./install.sh --doctor, do not fetch the Remotion best-practices skill, do not scaffold an episode, do not run selftest.sh, any library tool or npx remotion, and do not render anything: setting up the Standard or Full system, providing a storyboard and a script, and building an episode are my decisions, later, when I actually ask for them. 3. When I say "yes, install", install it as a skill: copy the whole papercut-build folder into the agent's skills directory, in a folder named "papercut-build", keeping its 1 file (SKILL.md at its root). A manual clone followed by copying the skills/papercut-build folder into ~/.claude/skills/ is the route described here; any other route is fine only as long as every installed file matches its fingerprint. 4. Verify with no network calls: frontmatter valid (name: papercut-build), the 1 file in place with a matching fingerprint. Do not run any library script or tool to verify. 5. Report what was installed, where, and what I still need to do myself (for example: get the whole system set up separately if it is not already — a Standard or Full install of the full repository through papercut-setup / install.sh, with Node 20+, npm, ffmpeg, rsync, Python 3 with numpy, Pillow and soundfile and Chrome or Chromium, knowing that setup fetches npm packages, the Remotion skills and, on the first render, Remotion's headless Chrome, and that Full additionally downloads a local voice model; load the Remotion best-practices skill; run the earlier stages so papercut/STORYBOARD.md and narration/script.json exist, since this skill builds from them; check the Remotion licence before commercial use; using the skill is a separate step that only starts when I ask to build or render, and the use-time sends disclosed above — npm i when no donor node_modules exists, the proof-page fetch to the URL I supply, the optional image-tool prompts, and the scene briefs to my harness's model provider when parallel subagents run — happen only then). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.