Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/skillmedev/creator-studio/video-hook-writernpx skills add SkillMedev/creator-studio --skill video-hook-writergit clone --depth 1 https://github.com/SkillMedev/creator-studioWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/skillmedev/creator-studio/video-hook-writer)<a href="https://agentmods.dev/skills/skillmedev/creator-studio/video-hook-writer"><img src="https://agentmods.dev/badge/skills/skillmedev/creator-studio/video-hook-writer.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00167 | $0.01536 |
| Opus 5 | $0.00084 | $0.00768 |
| Sonnet 5 | $0.00033 | $0.00307 |
| Haiku 4.5 | $0.00017 | $0.00154 |
Grade A, and why
Video Hook Writer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Video Hook Writer
On short-form video, the first three seconds are the entire audition: the platform decides distribution from early retention, and the viewer decides everything from the first frame and first spoken words. This skill produces hook combinations - what the viewer hears plus what the viewer sees - that arrest the scroll and open a question the video must answer. The costly mistake it prevents is a strong video that never gets watched because seconds one through three were setup instead of signal.
Operating procedure
Step 1: gather inputs
- The video's payoff: the one thing the viewer knows or can do at the end. If the creator cannot state it in one sentence, stop - the problem is upstream of the hook.
- Target platform: TikTok, Reels, or Shorts. Default to producing for the creator's primary platform and noting adjustments for the others.
- The creator's on-camera register (raw and conversational vs. polished and produced). If unknown, guess from their niche and label the guess.
- Any concrete assets in the content: numbers, timeframes, named results, a visible transformation. Hooks built on specifics beat hooks built on adjectives.
Step 2: write both channels of every hook
A hook has two components and both must deliver signal, not setup. The spoken line must be complete enough to create a question in the viewer's mind. The visual frame must be unexpected, specific, or slightly unresolved - because autoplay may be muted, frame one has to work silently. Specify both for every variant: never hand back a spoken line with no visual direction. Ban from frame one: black frames, slow zooms, logo stings, and a talking head with no context.
Step 3: generate variants across the six formulas
Produce at least three variants per video concept, each from a different formula, each labeled by formula name so the creator can A/B test and build pattern intuition. The highest-performing hook is almost never the first draft.
- Callout - addresses the viewer directly: "If you do X, stop right now."
- Curiosity Gap - "The reason most people fail at X is not what you think."
- Transformation Promise - "I went from X to Y in Z days - here's how."
- Controversy Opener - "Unpopular opinion: X is actually killing your Y." Only when the take is genuinely contested.
- List Tease - "Three things I wish I knew before starting X."
- Visual-First - the action or result is shown before a word is spoken; the caption or voiceover then names it.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 82 lines · 167 tokens per session scan A afcad8abaa77
Video Hook Writer is a skill published in the GitHub repository SkillMedev/creator-studio (5 stars, last pushed 2mo ago), licensed MIT. It adds 167 tokens to every session and 1,536 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
linkedin-message-writer
Research LinkedIn profiles and write personalized messages for any LinkedIn message type — connection requests, InMails, DMs, message requests, post comments, and comment replies. Takes LinkedIn URLs as input, researches each person (profile data + recent posts via Apify), and generates messages tailored to each…
generate-voice-guide
Generate a personal voice guide for X (Twitter) and/or LinkedIn by scanning a user's past posts and iteratively refining with sample-and-feedback loops. Produces a structured markdown voice guide that sibling skills (create-x-content, create-linkedin-content) consume to draft in-voice posts. Different from…
create-linkedin-content
Draft voice-tuned LinkedIn post variants from a free-form brief. Reads a personal voice guide (generated via generate-voice-guide), produces 2–5 variants with distinct framings, applies LinkedIn-specific defaults (arrow bullets, "why this matters" beat, 150–500 words), and self-checks against the voice guide's banned…
create-x-content
Draft voice-tuned X (Twitter) post variants from a free-form brief. Reads a personal voice guide (generated via generate-voice-guide), produces 2–5 variants with distinct framings (simple-howto, problem-first, hype, mechanism-breakdown, etc.), and self-checks against the voice guide's banned phrases before returning.
ecommerce-full-pipeline
电商运营在开展跨境电商或闲鱼捡漏业务时，若需解决选品难、上架繁琐等痛点，必用此技能！一键打通“爆品挖掘→1688采集→多平台上架→推广文案→短视频生成”全自动流水线，轻松实现端到端自动化，让开店运营效率翻倍。.
Read and post on X/Twitter from the CLI - fetch tweets, timelines, and search; post tweets, replies, threads, and media. Cookie-based auth via twikit, bootstrapped from the local Dia browser session, secrets stored in gopass. Auto-falls back to agent-browser when twikit hits internal-API drift or X's "looks automated"…