talking-head-recut

An agent skill by heygen-com, from heygen-com/hyperframes. Tags: automation, captions, design, social-listening, video.

What it does

Package an existing talking-head / interview / podcast video with timed, designed GRAPHIC OVERLAY cards — kinetic titles, lower-thirds, data callouts, quotes, side panels, picture-in-picture — synced to the transcript, on a 16:9 / 9:16 / 4:5 canvas of your choice; the clip plays untouched underneath. Trigger on "graphic overlays", "on-screen graphics", "package / dress up my video". Not plain subtitles (/embedded-captions). Unclear → /hyperframes.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add heygen-com/hyperframes --skill talking-head-recut

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/heygen-com/hyperframes
cp -r hyperframes/skills/talking-head-recut ~/.claude/skills/talking-head-recut

Safety box score

Grade C, moderate risk: 35 of 100 risk points, rated 2026-10-19. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill is a local video-processing and HTML-composition workflow: it probes media with `ffprobe`, extracts audio with `ffmpeg`, transcribes with local Whisper via `npx hyperframes`, then renders an MP4. The main risks are routine command execution plus some registry fetches, bounded file writes in the work tree, hidden operational instructions in comments, and an opaque bundled `gsap.min.js` renderer dependency.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	2 (moderate)	It runs real commands such as `npx hyperframes transcribe`, `npx hyperframes render`, `ffmpeg`, and `ffprobe`, so the agent executes local binaries as part of the workflow.
Network access	2 (moderate)	The skill uses `npx hyperframes` and explicitly asks for `npx hyperframes skills update talking-head-recut`, which can contact the npm registry even though transcription itself is local Whisper.
Filesystem reach	1 (low)	Writes are mostly bounded to `videos/<project-name>/` and `public/` (for `metadata.json`, `transcript.json`, `public/index.html`, `output.mp4`, etc.), with no stated writes outside the workspace.
Secrets & credentials	0 (none)	It says transcription needs no API key and does not request tokens, passwords, browser sessions, `.env` files, or keychain access.
Privilege & persistence	0 (none)	No `sudo`, global install, service registration, shell startup edits, or OS setting changes are requested.
Instruction hijack surface	2 (moderate)	The skill contains hidden operational instructions in HTML comments, e.g. `<!-- IMPORTANT: every card-host MUST carry BOTH "card-host" and "clip" classes. -->`, which is hidden text that steers the agent.
Irreversible actions	0 (none)	It automates local transcription/rendering but does not post, publish, deploy, send messages, pay, book, or delete anything.
Transparency	2 (moderate)	Key runtime assets are not reviewable here: `assets/vendor/gsap.min.js` is bundled but unread, and the shipped font `.woff2` files are binary and opaque.

Before installing

Read `SKILL.md` around the `npx hyperframes skills update ...` step and decide whether you accept registry fetches.
Inspect `assets/vendor/gsap.min.js` provenance/version before trusting the render path.
Run `npx hyperframes doctor` in a scratch workspace and confirm all writes stay under `videos/<project>/` and `public/`.
Skim the hidden HTML comments in `SKILL.md` if you want to avoid following embedded agent directives.

Rated 2026-10-19 by an automated review of 21 files from the skill's repository.

Source

Repository: heygen-com/hyperframes (all skills from this repository)
Path: skills/talking-head-recut/SKILL.md
Branch: main
Updated: 2026-09-19

captions-overlay — Overlay doctrine for the embedded-captions workflow — the caption MODEL (drop / rail / embed) and the rule that captions are an OVERLAY composited on top of.
embedded-captions — Add captions or subtitles to an existing single-subject talking-head video without editing the footage.
general-video — Author or edit a custom HyperFrames composition when no specialized workflow fits, or when BRIEF.md sets flow: companion. Use for longer or multi-scene pieces.
motion-graphics — A short, design-led motion graphic where motion is the message — kinetic typography, stat count-up, chart/data-viz hit, logo sting / brand lockup.
oversized-cursor — House-style oversized macOS cursor technique for HyperFrames launch videos.
product-launch-video — Turn a product or marketing URL, pasted script, or brief into a product launch / promo video — SaaS promos, feature reveals, product demos.