agentleFS
moses607moses607/socialforge/skills/hook-machine/SKILL.md

hook-machine

Skill3 starsChanged 4 months agoMITView on GitHub

Generate dozens of scroll-stopping hooks (the first 1-3 seconds / first line) engineered for a specific platform, audience, and format. Use when the user needs hooks, openers, titles, first lines, "make this stop the scroll", or their content gets impressions but no watch-time/clicks. Trigger on hooks, opening lines, titles, thumbnails-text, or low retention at the start. Works with any capable model.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Never ship the first hook.

    line 55
  2. Distinctive:

    No preamble. If the first 3 words don't stop a thumb, rewrite.

    line 56

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Hook Machine

The hook is 90% of the outcome. On every feed, the first 1-3 seconds (video) or first line (text) decides whether the algorithm keeps showing your content — because it measures whether humans stopped. A great piece with a weak hook dies; a mediocre piece with a great hook travels. So generate many hooks and select ruthlessly; never ship the first one.

The mechanism a hook must trigger

A scroll-stopping hook creates an open loop the brain needs to close. It does one of: promises a specific outcome, breaks a pattern/expectation, names a tension the viewer feels, or makes a claim they must verify. If it does none of these, it's a topic label, not a hook.

Method — generate wide, then cut

  1. Extract the payload. What's the single most valuable/surprising thing in this content? The hook sells that, not the topic.
  2. Run the hook archetypes — generate 3-5 hooks per archetype:
    • Contrarian — "Everything you've been told about X is wrong."
    • Specific result — "How I got 12,000 followers in 9 days with 3 posts."
    • Curiosity gap — "The one setting that doubled my reach (nobody talks about it)."
    • Callout — "If you post daily and still get 200 views, watch this."
    • Stakes/loss — "You're losing followers every time you do this."
    • Story cold-open — "I almost deleted this account. Then this happened."
    • Question they can't not answer — "Why do some 5-second clips get 2M views?"
    • Number/list — "5 hooks that print reach."
  3. Sharpen with specificity. Replace vague words with numbers, timeframes, names. "Grow fast" → "0 to 10k in 30 days." Specific = believable = clickable.
  4. Front-load. The stopping word must be in the first 3 words. Cut every warm-up syllable ("So today I want to talk about…" → delete).
  5. Score & cut. Rate each hook 1-5 on: stops-the-scroll, curiosity, specificity, truthfulness. Ship only 4s and 5s.

Output template

## Hooks for: <content> | Platform: <x> | Audience: <who>
Payload (what the hook sells): <the one valuable thing>

### Top 5 (ranked)
1. "<hook>" — archetype: <..> — why it works: <..>
...
### Bench (10-15 more, by archetype)
<grouped list>

### Paired visual/text-on-screen (first frame)
<3 on-screen text options for the first frame/thumbnail>

Platform variants

  • TikTok/Reels/Shorts: spoken hook + on-screen text hook must differ and stack; movement in frame 1.
  • YouTube (long): title + thumbnail text are the hook; they must create one loop together, not repeat.
  • X/Threads: first line only; the second line must earn the "show more" click.
  • LinkedIn: first 2 lines before "…more"; lead with a concrete result or a contrarian claim, never a preamble.

Rules

  • Never ship the first hook. Generate 15+, ship the best 1-3.
  • No preamble. If the first 3 words don't stop a thumb, rewrite.
  • Specific beats clever. A real number outperforms a pun.
  • The hook must be true to the content — a hook the payload doesn't pay off kills retention and trust (the algorithm punishes the drop-off).
  • Match the viewer's existing conversation in their head; don't announce your topic, enter their tension.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.