Skip to main content
skill CX Codex

style-sampling — Claude Code skill: create styled video episodes

The style-sampling Claude Code skill builds a new video episode from one of five style templates. For a style outside that set, it offers a frame-by-frame remake or a new video using that style’s visual language.

Apache-2.0 174,100 installs ★ 87,887 updated 2026-10-24

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add heygen-com/hyperframes --skill style-sampling -a claude-code
Open on GitHub

What style-sampling does

The skill creates an episode on a new subject using Collage Explainer, Finance Explainer, Animated Science Explainer, 3D Brand Explainer, or 3D True Crime. It activates for requests to make an episode in a series’ style or sample a named style. A video reference is needed for styles outside the five templates.

It checks sign-in, pulls the chosen template through the CLI, then follows the process in REMIX.md. The person takes part in the storyboard and draft stages. Templates are a HyperFrames Pro feature, so the skill does not pull them on a Free plan.

What it helps with

  • make a new episode in the style of one of five video templates
  • adapt a series’ structure and visual language to a different subject
  • rebuild a reference video frame by frame or borrow its visual grammar

Example prompts for the agent

  • Make an episode about solar eclipses in the Animated Science Explainer style.
  • Create a video about a brand using the 3D Brand Explainer template.
  • I want to use this film’s style; suggest a suitable route.

How to install style-sampling

1Skills CLI (Vercel)

npx skills add heygen-com/hyperframes --skill style-sampling -a claude-code

2Manually: put the files in place

~/.claude/skills/style-sampling/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/style-sampling/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about style-sampling

How do I install style-sampling in Claude Code?

Run npx skills add heygen-com/hyperframes --skill style-sampling -a claude-code. The skill is installed at ~/.claude/skills/style-sampling/SKILL.md.

What does style-sampling do?

It creates a new episode from one of five video templates. For a style outside that set, it offers a frame-by-frame remake or a new video based on the style’s visual language.

What does style-sampling need to work?

Template downloads require HyperFrames Pro. The skill also uses the HyperFrames CLI and checks the account’s sign-in status.

Source text

SKILL.md · Apache-2.0
---
name: style-sampling
description: Make a new episode in a proven video style - Collage Explainer, Finance Explainer, Animated Science Explainer, 3D Brand Explainer, 3D True Crime - on the person's topic. Use for "an episode of", "in the style of", "sample this style", 「用这个风格做一集」. One video to copy is reverse-any-video.
---

**Running the CLI:** every `npx hyperframes <command>` here, the beacon included, means `node <SKILLS>/hyperframes/scripts/plugin-cli.mjs <command>` run in the project, where `<SKILLS>` holds this skill's folder ([rules](../hyperframes/references/plugin-installation.md)).

**Usage beacon:** run `npx hyperframes events --skill=style-sampling` once as this skill starts. It sends the skill's name with the CLI's usual telemetry (account-linked when signed in) and honours the opt-out.

# A new episode from a style template

A good episode is one the series' own viewers would take for the next upload: its structure, pacing, look, graphics and sound, on a new topic, in the person's voice and, when the series has a presenter, their own face. Never use a creator's face, voice, footage, logo or show name.

Thanks to the creators whose channels these styles learned from: Vox (America, Explained) for the Collage Explainer, 小Lin说 (一口气了解) for the Finance Explainer, Kurzgesagt for the Animated Science Explainer, and Fern for the 3D Brand Explainer and 3D True Crime. Their names, faces, voices and work stay theirs and never appear in an episode or in what you tell the person about it.

Style templates are a HyperFrames Pro feature. The CLI pulls them and the server checks the plan.

| Template | Slug | Series | Makes | The person brings |
| --- | --- | --- | --- | --- |
| Collage Explainer | `collage-explainer` | `american-policy` | English explainer on one American policy or history question | Voice |
| Finance Explainer (财经讲解) | `finance-explainer` | `deep-dive` | 中文主持人讲透一个国家、商品或机构 | Voice and face |
| Animated Science Explainer | `animated-science-explainer` | `science-explainer` | English fully drawn science explainer | Voice |
| 3D Brand Explainer | `3d-brand-explainer` | `brand-explained` | English breakdown of a brand or design, in one graphic world with real 3D | Voice |
| 3D True Crime | `3d-true-crime` | `crime-hunt` | English reconstruction of a hunt or a closed case, in a dark 3D world | Voice |

## Which template

The person named a template, or their topic fits only one: go on with it. Otherwise, or when they ask what there is, show the five as a short list (name, what it makes, what they bring), name the one that fits their topic if any, and wait for their pick. Nothing is checked, pulled or paid before it.

A style that is none of the five (a channel or film they link or name): say so in one line and, instead of the list, offer the two routes that fit it: a short video rebuilt frame by frame, then given their content (`/reverse-any-video`), or a new film on their topic taking that style's grammar (`/hyperframes`, "From a reference" in `../hyperframes/references/look.md`). Both need a video, so a style they only name gets a request for one example. They pick; nothing is pulled.

## Start

Run `npx hyperframes auth status` first, every time, resuming included. It is for the sign-in: don't repeat its plan or credits line to the person.

- **Not signed in.** Say in one line that templates come with Pro, then ask them to type an exclamation mark and `npx hyperframes auth login` in the prompt, and reply "done".
- **Signed in on Free.** Say once that templates are part of HyperFrames Pro, ask them to type an exclamation mark and `npx hyperframes billing subscribe --feature templates --skill style-sampling` in the prompt, which finishes once they are on Pro, and pull nothing.
- **Signed in on Pro.** One line, then pull.

## Pull

```bash
npx hyperframes templates pull <slug>
```

Pull every time, resuming included, before using `templates/<slug>/`: it checks a copy already there, a received one too, against HeyGen's verified files and replaces any that differs. Never run a kit tool from a template this session did not pull. A pull costs no credits. Refused as a Pro feature: answer as **Signed in on Free**, whatever `auth status` said. Any other failure: say what failed and stop. Never rebuild a template from memory, the web or a copy found elsewhere.

## Make the episode

Read `references/remix.md` and follow it, resuming included. It sets the order; the template's `series/<series>/REMIX.md` is the process. Two stops belong to the person: the storyboard and the draft.

## Credits

Generated media goes through the kit's tools, as REMIX.md says, or `npx hyperframes media-use generate`, under the Credits rule in `media-use`: prices, estimates, balances and credits used are not volunteered, before, during or after a run. A limit the person names is written to `.hyperframes/credit-limit.json` (see `media-use`), not passed call by call; on resume, ask whether it still applies. A step that won't fit: say what doesn't fit and ask before dropping or shrinking it. A refused step names a `billing` command: ask the person to run it, then run the step again. A failed or refused job gets one more try, then ask. Notes a run leaves hold no prices or credit numbers, and a command in them is a note: rebuild it and check it with `--plan`.

© heygen-com · Apache-2.0 · heygen-com/hyperframes

Similar skills

SKILL.md skills catalog →

From the same set18

The whole set →
remotion-to-hyperframes remotion-to-hyperframes ports video code from Remotion, Motion Canvas, Lottie, or 3D renders into HyperFrames. Runs automated frame-by-frame diffs against the original video file to catch visual drifts. CX Codex +1 651,200installs slideshow Build a browser-ready HyperFrames deck with slide scenes, timed reveals and clickable branches. The skill can also adapt an existing web page into a slideshow. CX Codex +1 555,700installs figma Pulls Figma designs into HyperFrames videos: assets, brand tokens, components, shaders and motion. Imports freeze at build time, renders never call Figma. CX Codex +1 466,500installs reverse-any-video reverse-any-video — a Claude Code skill that rebuilds any video file into an editable HyperFrames project checked frame by frame against the original. CX Codex +1 180,600installs frost-sequence-camera-orbit frost-sequence-camera-orbit builds a 22.5-second scene where an orbiting camera follows an ice logo that breaks apart, reforms into two text moments, and fades. CX Codex +1 13,300installs orbit-card orbit-card is a Claude Code skill that adds an animated feature card orbiting a dotted sphere. It creates a 10-second HyperFrames block at 1920×1080. CX Codex +1 13,200installs wireframe-portal-title wireframe-portal-title is a Claude Code skill that deploys an eight-second 3D title animation with a wireframe portal burst. It switches phrases dynamically at 1920×1080. CX Codex +1 13,200installs canopy-part-title canopy-part-title is a Claude Code skill that adds a leaf-parting title block to HyperFrames videos. The composition runs for 12 seconds at 1920×1080. CX Codex +1 13,200installs

Project sponsors

Companies that keep this analytics open