nateherkai/hyperframes-student-kit

HyperFrames Student Kit – AI‑assisted video‑editing playground

What it is – A reusable, open‑source toolkit that lets you edit your own footage with the help of large‑language‑model agents (Codex or Claude). The repo bundles a collection of “skills” (agent‑driven commands), motion‑graphics assets, scene templates, and a set of helper scripts that automate common post‑production tasks such as silence removal, mistake detection, transcript‑based retiming, beat‑sync validation and short‑form reel creation.

Why it matters – It demonstrates a concrete workflow where an LLM acts as a video‑editing assistant, turning natural‑language prompts into concrete FFmpeg/GSAP operations. The kit is designed for students and creators who want to experiment with AI‑driven editing without building the infrastructure from scratch.

---

Core components

Component	What it provides
14 “skills”	Pre‑written agent prompts for tasks like edit-video, cut-silences, cut-mistakes, short-form-edit, each with helper scripts and reference material.
406 draft motion‑graphics cards	Ready‑to‑use GSAP‑based graphics (two visual styles) with CSS tokens and editable slots; metadata lives in style-library/registry.json.
Scene templates	Two full‑scene layouts – a dark graph‑paper background and a left‑glass pop‑out – that can be combined with the cards.
Transcription & cutting pipeline	Optional integrations with ElevenLabs Scribe, OpenAI Whisper, or local Whisper for word‑level transcripts; scripts to cut dead air, flag mistakes, and generate an edit‑decision list (EDL).
Short‑form workflow	End‑to‑end commands for turning a long recording into a 9:16 Reel or YouTube Short, including hook/payoff planning, caption timing validation, B‑roll insertion and sound‑design hints.
Demo & teaching projects	Twelve example student projects and a synthetic starter composition that run without any external API keys.

---

Who it’s for

• Students and educators learning how to combine LLM agents with media pipelines.
• Content creators who want a scriptable, AI‑guided way to trim and polish footage.
• Developers interested in building or extending AI‑agent‑driven video tools (the repo is a ready‑made reference implementation).

---

Getting started (from the README)

1. Prerequisites – Node 22+, Git, FFmpeg (with ffprobe), and Chrome/Chromium on your PATH.
2. Clone & install

   git clone https://github.com/nateherkai/hyperframes-student-kit.git
   cd hyperframes-student-kit
   npm ci               # installs dependencies
   npm run setup        # checks tools, creates .env template
   npm test             # sanity‑check
   

3. Run the demo

   npm run demo
   cd video-projects/demo
   npx hyperframes lint
   npx hyperframes preview   # watch the 8‑second animation
   npx hyperframes render --quality draft --output renders/demo.mp4
   

4. Edit your own footage – Open the folder in Codex or Claude Code and issue a natural‑language command, e.g.

   $edit-video  /path/to/my‑recording.mp4
   

   The assistant will invoke the appropriate skill (cut-silences, cut-mistakes, etc.) and produce a draft edit.
5. Create a short‑form Reel – Use the short-form-edit skill with a prompt like:

   /short-form-edit  /path/to/lecture.mp4  make a 9:16 Reel with hook, captions and beat‑sync
   

   The workflow validates caption timing, scene coverage and footage reuse before rendering.

---

Optional AI services

• ElevenLabs Scribe – cloud transcription (default helper).
• Kie.ai – generates supplemental video/image assets.
• OpenAI Whisper or local Whisper – alternative transcription back‑ends. You only need to supply API keys for the services you choose; the synthetic starter works entirely offline.

---

Limitations & considerations

• The kit relies on external APIs for high‑quality transcription or asset generation; without them you get a purely local workflow that skips those steps.
• Motion‑graphics cards and templates are draft assets – they are meant to be customized (CSS tokens, text slots) before final production.
• The assistant commands ($edit-video, /short-form-edit, etc.) are tied to the Codex and Claude Code IDE extensions; using them outside those environments requires manual invocation of the underlying scripts.
• Rendering quality is controlled by the --quality flag; higher‑quality renders will take longer and may need more system resources.

---

Quick reference links (from the repo)

• Agent instruction reference – AGENTS.md
• Motion‑graphics vocabulary – MOTION_PHILOSOPHY.md
• Style‑library guide – style-library/GUIDE.md
• Workflow walkthrough – docs/WORKFLOW.md
• Prompt recipes – docs/PROMPTS.md
• Short‑form editing guide – docs/SHORT-FORM.md
• Tool & API‑key setup – docs/TOOLS-AND-API-KEYS.md

---

Bottom line – HyperFrames Student Kit is a hands‑on, open‑source sandbox for exploring how LLM agents can drive video‑editing pipelines, complete with assets, scripts, and example projects that let you go from raw footage to polished short‑form content with only a few natural‑language commands.