PRO NEW video v1.6.2 HyperFrames

Caption Studio

Word-by-word captions that carry a talking-head clip.

Turns a raw talking-head recording into a fast, watchable vertical clip. It transcribes the audio, styles word-by-word captions, cuts filler and dead air, and adds b-roll or quote cards where the story needs a visual break. Every clip is delivered in 9:16 first, with a 16:9 version for YouTube and LinkedIn.

9:1616:9 22k installs 4 of 5 agents tested

Preview format

Platform

Safe zone preset approximate platform UI

REC 00:15 9:16

Caption Studio

Rendered by this skill · unedited

Video preview

Placeholder clips, not real skill output. The 9:16 view shows approximate platform safe zones — always check the final render in the app.

Takes

Example commands for this skill. Copy one into your agent.

3 takes

Take 01 Full run Use caption-studio on this interview clip, keep it under 45 seconds, and make a punchy 9:16 cut with keyword highlights.
Take 02 Vertical test Use caption-studio to make a 5-second 9:16 test run.
Take 03 Landscape master Use caption-studio to render the 16:9 master from the same project, with no copy changes.

Install

Three ways in. Same skill.

v1.6.2

Browser sign-in on first install Commands copy now; sign in when you run it.

My agent

Installs into ~/.claude/skills for Claude Code.

Paste this into your agent. It reads the skill's install doc, sets it up and runs a vertical test.

Prompt

Read https://skillnai.com/skills/caption-studio/install.md and install it for me, then do a 9:16 test run.

Add skill

npx skillnai add caption-studio --agent claude-code

Doctor

npx skillnai doctor caption-studio

Run this in Claude Code.

MCP setup

claude mcp add --transport http skillnai https://mcp.skillnai.com

MCP commands are illustrative placeholders. Never paste credentials into a prompt.

Get Pro Buy this skill

Browser sign-in happens on first install of this premium skill. Never share passwords, API keys or credentials with an agent.

What you get

Everything the skill produces from a single run.

15–60s 1080×19201920×1080

9:16 clip with word-by-word captions
16:9 cut for YouTube and LinkedIn
B-roll, quote-card and emphasis inserts
Editable .srt and caption style presets

Input

A talking-head video (MP4 or MOV), plus an optional brand kit or transcript.

Good fit / Not for

Know before you run it.

Good fit

Founder and expert talking-head clips
Podcast clips for Reels and Shorts
Course and tutorial snippets
Turning one recording into a week of posts

Not for

Multi-person interviews with heavy crosstalk
Clips where captions must match a strict legal script
Footage with unusable audio
Full podcast episodes

Requirements

Run the doctor command after install to verify the environment.

Tool	Version	Status	Purpose
`node`	>=20	Required	Runs the skill's build scripts and render pipeline.
`ffmpeg`	any	Required	Extracts audio, trims clips and encodes the final video.
`chrome`	any	Required	Headless Chrome renders the caption composition.
`FAL_KEY`	any	Optional	Optional generated b-roll and music via FAL.

External costs

Rendering is local and free. Optional FAL jobs are billed by FAL per job; the skill lists costs and waits for approval.

The skill shows an estimate and waits for approval before any paid job starts.

Tested versions

We only mark an agent as tested after a clean install and render.

Tested model: Claude Sonnet 4.5
Last tested: 2026-10-06
Default format: 9:16
Engine: HyperFrames

Agent	Status	Install directory	Model
Claude Code	Tested	~/.claude/skills	Claude Sonnet 4.5
Cursor	Tested	~/.cursor/skills	Claude Sonnet 4.5
Codex	Tested	~/.agents/skills	Claude Sonnet 4.5
Gemini CLI	Untested	~/.gemini/skills	—
OpenCode	Tested	~/.agents/skills	Claude Sonnet 4.5

Render command

Run the same prompt to reproduce this preview. Output is rendered by the skill, unedited.

prompt · seed 0421

Use caption-studio on this interview clip, keep it under 45 seconds, and make a punchy 9:16 cut with keyword highlights.

Workflow

What the agent does, step by step.

01
Transcribe the recording and detect speakers, pauses and filler words.
02
Pick the strongest 15–60 second segment or follow your time range.
03
Style word-by-word captions with a safe-zone-aware layout.
04
Place b-roll and quote cards over low-energy moments.
05
Render the 9:16 master, then re-frame a 16:9 version.
06
Verify caption timing, contrast and platform UI safe zones.

Platform presets

Each skill renders the formats its target platforms expect.

Default 9:16

TikTok

Full-screen vertical with safe zones.

9:16 Vertical

Instagram Reels

Vertical with caption-safe framing.

9:16 Vertical

YouTube Shorts

Vertical with top and bottom clearance.

9:16 Vertical

Instagram

Square feed and carousel covers.

9:16 Vertical

YouTube

Landscape master and thumbnails.

16:9 Landscape

Feed video and document posts.

16:9 Landscape

All formats in this skill: 9:16 1080×1920 16:9 1920×1080

Changelog

One entry, derived from the version and the last tested date.

v1.6.2 2026-10-06

v1.6.2 — re-tested with Claude Sonnet 4.5

Verified the install prompt, the 9:16 / 16:9 exports and the safe-zone check on Claude Code, Cursor, Codex and OpenCode. Last run on 2026-10-06.