faceless-explainerby HeyGen

Faceless Explainer Skill.

A Claude Code and Codex skill for motion and video: turns text, notes, or an article into a faceless explainer video with invented scene visuals.

Motion & Video
Video

Download OpenDesign View usage guide

heygen-com/hyperframes · skills/faceless-explainer/SKILL.mdSKILL.md

--- name: faceless-explainer description: Turn arbitrary text — an article, notes, a topic, a brief — into a faceless explainer video: there is no site or footage to capture, so the visuals are invented per scene (typography, abstract graphics, diagrams, data-viz). Use for topic explainers, concept breakdowns, how-tos, listicles. Not a video built from a website (/product-launch-video — promo or tour). Unclear → /hyperframes. --- ## Step 0: Setup ## Step 1: Brief (no capture) ## Step 2: Design System ## Step 3: Storyboard and Script ## Step 3.1: Audio ## Step 4: Frame Visual Design

Front matter and section headings read from the upstream file on 2026-09-14 · 29,476 characters in full

01 What it does

What the faceless-explainer Skill Does in Claude Code and Codex

Summary, workflow and outputs, all read from the skill's SKILL.md.

From the SKILL.mdRead 2026-09-14 · not a recorded run

In the Skill's Own Terms

Turns arbitrary text — an article, notes, a topic, a brief — into a faceless explainer video with invented scene visuals (typography, abstract graphics, diagrams, data-viz). Works through seven orchestrated steps: setup with brief confirmation, synthetic capture package (no website), design system selection from shipped frame presets with optional brand token remixing, storyboard and script using narrative design to reshape input, audio generation with TTS and BGM, visual design adding time-coded shot sequences with invented elements, parallel frame building via sub-agents, and final render.

User-gated checkpoints at Steps 0, 3, and 6 in collaborative mode. Faceless means no capture step and no real asset inventory — every visual is invented downstream by workers.

What it produces

STORYBOARD.md with frame-by-frame teaching plan, time-coded shot sequences, and invented focal/roles
compositions/frames/NN-*.html — one HTML composition per frame built by sub-agents
renders/video.mp4 — the final rendered video in landscape 1920x1080, portrait 1080x1920, or square 1080x1080

See the example prompt ↗

How It Works

01
Confirm Brief and Initialize Project
BRIEF.md exists and determines the mode — read it and ask nothing. If missing and project is fresh, run the intent layer to lock the brief.
02
Create Synthetic Capture Package
Save the user's full input verbatim as capture/extracted/visible-text.txt (the source of information).
03
Choose Frame Preset and Build frame.md
Pick one shipped preset from hyperframes-creative/frame-presets/ that fits the topic and tone.
04
Write Storyboard with Narrative Design
Turn text into frame-by-frame teaching plan using story-design.md structure.
05
Generate Audio and Add Visual Design
Run audio.mjs in background for TTS, word timings, BGM from HeyGen library.
06
Build Frames and Assemble Index
Sync durations, build per-frame packets, dispatch one sub-agent per frame in parallel. Each worker writes compositions/frames/NN-*.html. Build captions in background.

02 Find your fit

When to Use the faceless-explainer Skill

Use the faceless-explainer skill in Claude Code or Codex when the job is motion and video of this kind. Fit and limits below are taken from the skill's own file.

Good Fit

Topic explainers, concept breakdowns, how-tos, and listicles
Explaining a topic from text with no product and no website to capture
When every visual needs to be invented: typography, abstract graphics, diagrams, or data-viz

Know the Boundaries

Video built from a website for product promo or tour
When there is a capture step or real asset inventory needed
Bare "make a video" request or any uncertainty — must go through intent layer first

What to Provide

Arbitrary text sourceAn article, notes, a topic, or a brief — the full input saved verbatim as capture/extracted/visible-text.txt.
Optional user scriptIf the user pasted a script or wants their wording kept, save it verbatim as user_script.txt. VO_MODE (verbatim or restructured) comes from BRIEF.md.
Optional brand colors or fontsLeave tokens.json colors and fonts empty unless the user explicitly gave brand colors or fonts.
Optional voice preferenceIf the request named a voice, gender, or tone, pick a matching voice id and pass --voice <id>.
Optional real imageIf the user supplied a real image, place it under public/<basename> and note it for Step 3.

03 Inside the skill

Rules the faceless-explainer Skill Gives Claude Code and Codex

7 concrete instructions, defaults and limits the SKILL.md sets for the agent — the part of the file that changes the result.

Opening Rule for Brief Handling
If BRIEF.md exists, read it and ask nothing — the brief is settled.
No Capture Step for Faceless Videos
Do not run npx hyperframes capture — there is no URL. Do not create asset-descriptions.md or populate capture/assets/. Faceless visuals are invented in Steps 4-5, not captured.
Run Steps in Order with Gates
Work through Step 0 setup, Step 1 brief, Step 2 design system, Step 3 storyboard, Step 3.1 audio, Step 4 visual design, Step 5 frames, Step 6 finalize.
Leave Asset_candidates Empty by Default
Frames do not carry an asset inventory in faceless videos. Leave asset_candidates empty unless the user supplied a real public/<basename> image.
Search Catalog Before Inventing Looks
For every named look, effect, or transition the brief mentions, run npx hyperframes catalog --query with plain English before writing it into STORYBOARD.md.
Full-bleed Backgrounds on Clip Layer
A frame's ground (color field, gradient, grid) is its own full-duration background clip with class="clip".
Duration Sync Is Mechanical
Real voice duration wins; silent frames keep estimates. Run audio.mjs sync-durations after Step 3.1 finishes. Never hand-edit synced durations.

04 Put it to work

Install faceless-explainer in Claude Code or Codex

One npx skills add command, then a first task in your agent and a checklist for the result.

Add faceless-explainer to Claude Code, Codex or Your Agent

Run in your project; the installer asks which agent to add it to.

Terminal

npx skills add https://github.com/heygen-com/hyperframes --skill faceless-explainer

Generic skills CLI form for skills/faceless-explainer/SKILL.md; the repository may document its own path.

Give the Skill a First Motion and Video Task

Written in the skill's own terms; replace the bracketed parts with your material.

Starter prompt

Turn this article about compound interest into a 60-second explainer video for young adults. Use clean modern graphics with invented diagrams showing how money grows over time. Keep the tone approachable and use a female voice for narration. Format: [landscape/portrait/square]. Destination: [Instagram/YouTube/LinkedIn].

Source: written from the SKILL.md in the skill's own terms.

Check the First Result

What the SKILL.md itself says a correct result looks like:

Run npx hyperframes lint to validate structure
Run npx hyperframes check to catch text overflow and layout issues
Run npx hyperframes snapshot at frame midpoints to create contact sheet, glance at it for obvious breaks
Ignore 1-4px text_box_overflow on caption words (#caption-word-*, .caption-line) — expected false positive from snug line-height
Only act on text_box_overflow when it names a frame element (#el-NN-*), not a caption word
Run transitions.mjs inject and verify before final render to check transition timing

In OpenDesign import faceless-explainer from its SKILL.md link: Plugins → Add → Skill → Import from link, then pick it in the top bar.

Download OpenDesign Read the quickstart

05 Open source, traceable

Source, License and Verification

Where this page's facts come from, and when they were last checked.

Author: HeyGen
Repository: github.com/heygen-com/hyperframes Branch main
Skill file: skills/faceless-explainer/SKILL.md Content checksum recorded at verification; commit 558c11b
License: Apache-2.0 From LICENSE.
Repository stars: 49,761 GitHub snapshot taken 2026-09-14 for the whole heygen-com/hyperframes repository, which contains 9 skills in this catalog. Not a rating or usage count for faceless-explainer.
Verified by OpenDesign: 2026-09-14 Source file, path, license text and star count were checked. Not bundled in OpenDesign main (checked 2026-08-28); a live run has not been recorded on this page.

Inside the SKILL.md

Step 0: SetupConfirm brief, initialize project, write BRIEF.md, record preferences, show sign-in status
Step 1: Brief (no capture)Save input as visible-text.txt, create synthetic tokens.json, no website capture or assets
Step 2: Design SystemChoose shipped frame preset, run build-frame.mjs to remix brand tokens onto preset
Step 3: Storyboard and ScriptWrite frame-by-frame teaching plan with narrative design, present as proposal, loop until approved
Step 3.1: AudioGenerate narration with TTS, word timings, BGM from HeyGen library, audio metadata
Step 4: Frame Visual DesignAdd time-coded shot sequences, invented focal/roles, layout and motion inline per Scene
Step 5: Build FramesSync durations, build packets, dispatch sub-agents per frame, assemble index, build captions
Step 6: FinalizeInject transitions, run lint/check/snapshot, pause for review, render final MP4
Quick ReferenceFormats, faceless deltas vs captured-asset workflow, background scripts, reference table

The Skill's Own Description

“Turn arbitrary text — an article, notes, a topic, a brief — into a faceless explainer video: there is no site or footage to capture, so the visuals are invented per scene (typography, abstract graphics, diagrams, data-viz). Use for topic explainers, concept breakdowns, how-tos, listicles. Not a video built from a website (/product-launch-video — promo or tour). Unclear → /hyperframes.”

Front-matter description from SKILL.md. The full file is about 29,476 characters. Read the full file on GitHub.

06 Before you install

Questions About the faceless-explainer Skill

Answers come from the SKILL.md as read on 2026-09-14, not from a recorded run.

Does this work in collaborative and autonomous modes?

Yes. In collaborative mode, user approval gates run at Steps 0, 3, and 6 with sketch review at Step 4. In autonomous mode, the same summary posts as a heads-up and sketches collapse into the build, with one preview question kept at Step 6.

What if I don't have a HeyGen account for voice or music?

Step 0 shows sign-in status before proceeding. When signed out, audio falls back automatically to local engines (Kokoro for TTS). You can choose collaborative mode to wait for sign-in or autonomous to continue through available local engines.

What video formats does the output support?

Landscape 1920x1080, portrait 1080x1920, or square 1080x1080. The format is derived from the destination in BRIEF.md and set once in the storyboard frontmatter.

Faceless Explainer

Run faceless-explainer on a Real Motion and Video Task.

Download OpenDesign, import faceless-explainer from its SKILL.md link, paste the prompt above, and read the plan before the result.

Download OpenDesign Browse all skills