The hook research problem
Every creator I know has the same loop. You sit down to record. You don't know what to talk about. So you open Instagram. You scroll for 30 minutes "researching." You save 4 reels you'll never look at again. You give up and post something mid.
Hook research is the single biggest time-suck in content creation. And almost everyone does it manually.
The thing is — the data is sitting right there. Your competitors are publishing every day. Some of those posts are doing 10x their normal reach. Those are the hooks worth stealing. The problem isn't access. It's that nobody has time to find them, transcribe them, and turn them into something they can actually use.
What /hook-intel actually does
One command. It does six things automatically:
You hit /hook-intel refresh. You walk away. You come back to a fresh batch of proven hooks pulled from the only creators that matter — the ones already winning in your niche.
The 6-step pipeline
What's actually happening under the hood when you run it:
Why "outliers" — not just "most viewed"
This is the part most people miss when they try to do this manually.
A post with 100k views isn't impressive on a 1M-follower account. It's a flop. A post with 50k views from a creator who normally gets 5k? That's the one to study. The skill calculates each creator's median view count over their last 50 posts, then only flags content that beat that baseline by 5x or more.
You get the actual outliers. Not vanity numbers.
What you end up with
After a few weeks of this running on autopilot, you have:
- A live databaseof every hook that's outperformed in your niche.
- Reusable templates you can swipe for any topic — not word-for-word copies.
- Filters so you can pull only carousel-friendly hooks, video hooks, or specific creators.
- A connected /script command that automatically writes in the patterns proven to work.
I run this on 21 competitors. 96 hooks in the database. 27 templates. The whole thing refreshes itself every 7 days via cron — I don't even have to remember to run it.
Install it in 30 seconds
Just tell Claude Code to install this skill from the repo below. It'll handle everything — file paths, the setup wizard, the database schema.
Tell Claude: "install this skill for me: https://github.com/tenfoldmarc/hook-intel-skill"
Prefer to do it manually? Paste this into your terminal:
Then open Claude Code and type /hook-intel to start the setup wizard.
What you'll need
Three things. All free or close to it.
- Claude Code. The brain that runs the skill.
- An Apify account.Pay-per-use scraping. Costs pennies for the volume you'll use.
- A Supabase project. Free tier is more than enough.
The setup wizard walks you through wiring all three. You do it once. It runs forever.
When to skip this
If you're posting once a month, this is overkill. The setup pays for itself when you're publishing weekly or daily and burning real hours on research.
And if you don't have competitors you respect, build that list first. The skill is only as good as who you point it at. Pick 15-25 creators in your niche who are clearly winning. That's the input that matters.
The real shift
The mental shift here isn't about scraping. It's about treating content research as something a machine should be doing in the background — not something you do manually every Sunday night.
You should never be sitting there wondering what's working. The data is public. The tools are free. The only reason most creators aren't using this is that nobody told them they could.