---
name: hook-lab
description: Generate 10 scroll-stopping hooks for a topic — 10 directions, each privately scored 1–100. You pick your favourite blind, then the scores are revealed, then you choose — and get a focused strategy brief for the hook you picked. Reads _creator-profile.md for voice/audience. Triggers: "hooks for", "write hooks", "hook ideas", or after content-calendar.
---
This is a real Claude Code skill — the actual
hook-labfrom my Content-to-Cash pack, given away free. It's not a prompt or a swipe file. Drop it into Claude Code and it writes 10 hooks across proven frameworks, scores each one 1–100 for how hard it stops your audience, and hands you a strategy brief for the one you pick — why it works, who it lands on, and the CTA that ties it to your offer. Built by @buildwith.conrad.
A hook is the first line — the one that decides whether someone stops or keeps scrolling. On a Reel it has about 1.7 seconds to land; on X it's the first line; on a carousel it's the cover. Most first lines are generic ("5 tips to write better content") and the scroll just keeps moving. This skill writes ten genuinely different hooks for your topic — each from a different proven angle (curiosity gap, contrarian, transformation, and more) — shows them to you blind first so you trust your gut, then reveals a 1–100 score for each so you can see which one actually stops people. You pick one, and it hands you a short plan: why that hook works, who it lands on, the content to build around it, and the call-to-action that points at what you sell. No marketing background needed.
Claude Code (recommended):
- In your project, create the folder
.claude/skills/hook-lab/(make.claude/skills/if it doesn't exist). - Save this file as
.claude/skills/hook-lab/SKILL.md. - Open Claude Code in that project and type:
hooks for [your topic]. That's it — the skill takes over.
No project / just trying it: paste everything below the ── SKILL ── line into any fresh Claude chat,
then send your topic in one line. It works as a standalone too.
This skill is conversation-only — it writes no files. The hooks and the brief live in the chat. If you have a
_creator-profile.md(from the full pack'screator-onboard) in the folder above, it reads your audience, platform, voice and offer from it so every hook fits. No profile? It just asks you once for your audience + platform + one past post that did well, then runs — nothing else to set up.
Ten hooks, ten directions, one clean list. You pick the one you stop on longest — then the 1–100 scores are revealed — then you choose, and get a focused strategy brief for the hook you picked. The strategic work is done up front while the hooks are generated; only the decision is yours. Conversation + file read/write only.
- The creator says "hooks for X" / "write hooks" / "hook ideas", OR they're filling a calendar slot.
- Reads: ../_creator-profile.md (Audience, Platform(s), Voice, sample posts, Offer).
- Feeds: carousel-writer, caption-cta. Text/copy only — the hook is a first line, no video/script here.
- Topic — if the creator has none, don't just ask: offer 3 angles drawn from their niche + offer (or pull one from content-calendar) and let them pick. Never leave a beginner on a blank page.
- ../_creator-profile.md — pull Audience, Platform(s), Voice, sample posts, Offer. If no profile exists, ask once for audience + platform + one past post that did well.
- GET TOPIC. Get the topic; if none, offer 3 from niche + offer (or content-calendar) and let them pick.
- GROUND (internal, ~1 line back). Read the profile for audience, platform, and voice so every hook fits — first line lands in ~1.7s on a Reel, stands alone on X, is the cover on a carousel. Don't make them answer a calibration quiz; infer, state your read in one short line they can correct.
- GENERATE + PRE-PLAN (do the work here, show none of it). Write 10 hooks, each a genuinely different angle — no two the same move. Pull from these directions (skip any that force a weak line, swap in another): curiosity gap · contrarian · identity/mirror · loss/cost · concrete number · transformation before→after · shared-pain relatability · mistake/warning · bold promise · pattern interrupt. Bias 2–3 toward the offer's core transformation. Then, privately and for EVERY hook, prepare two things so they're ready whichever one wins: (a) a 1–100 score for how hard it stops THIS audience on THIS platform — honest, SPREAD (no clustering at 80–95); (b) a one-paragraph strategy brief — the angle, who it lands on + their awareness stage, the content angle to build it into, and the CTA that ties it to the offer. Voice-match; cut hype; never fabricate a stat (
[insert real number]). Reveal NONE of this yet. - PHASE 1 — BLIND LIST. One short line first — a hook is the first line that stops the scroll, not the whole post. Then the 10 hooks as a clean numbered list: just the lines. No scores, no labels, no strategy. Ask: "Which one do you stop on longest? Pick a number — before I show my scores." Wait for the pick. If they truly can't choose, let them say "you pick" and go with your top-scored one — never leave them stuck on the choice.
- PHASE 2 — REVEAL SCORES. Show all 10 ranked by your 1–100 score, each with a 1–3 word direction tag + a ≤6-word reason on the top 3. Call out where THEIR pick landed and which is your #1. Flag any
[insert real number]. Lean — a ranked list, not the report yet. - PHASE 3 — DECIDE (human-driven). Ask which they're running with — their gut pick, your top, or a blend. Wait for the decision. Do not pre-empt it; if they're unsure, recommend the higher-scored of (their pick vs your #1) and let them confirm.
- STRATEGY BRIEF — for the chosen hook only. Surface the brief you already prepared in Step 3 for the hook they picked: why it scores where it does, who it lands on (in plain words — "people who feel broke but don't know where it goes", never "problem-aware"), the content angle to develop it into, and the CTA that ties it to the offer. Then hand off — carousel-writer (a deck) or caption-cta (a single post). Copy/strategy only — no script, no rendering. A tight battle plan, not an essay.
A staged chat exchange (this skill saves no file — the hooks live in the chat):
- Phase 1: one short "a hook is the first line" note, then a clean numbered list of 10 hooks (NO scores, NO labels, NO strategy), then "which do you stop on longest? pick before I show my scores."
- Phase 2 (after they pick): the 10 ranked by 1–100, a 1–3 word direction tag each, a ≤6-word reason on the top 3, a note on where their pick landed + your #1, any
[insert real number]flag. - Phase 3 (after they decide): the chosen hook + its strategy brief (why it scores there · who it lands on · the content angle to build · the CTA/offer tie) + the one-line handoff to carousel-writer / caption-cta.
- Exactly 10 hooks, 10 genuinely different directions — no two the same move, no filler to hit the count.
- Phase 1 is BLIND: zero scores, labels, or strategy shown until the creator has picked. Instinct first, scores second.
- The strategy brief was PREPARED during generation (Step 3), not improvised after, and is delivered ONLY for the chosen hook — specific to it, never a generic dump.
- Scores are spread and honest (not all 80–95) and reflect THIS audience + platform.
- Voice-matched; no hype/AI-slop; every number real or flagged
[insert real number]. - Ends with the creator choosing ONE hook + its brief + the handoff to carousel-writer / caption-cta.
hook-lab is one skill in my Content-to-Cash pack — 8 skills that turn your reach into revenue,
using Claude Code alone:
creator-onboard → content-calendar → hook-lab → carousel-writer / caption-cta → content-repurpose → content-to-offer / list-engine
Each one hands off to the next through a shared creator profile, so the whole thing stays coherent instead of eight disconnected prompts — plan a week, write the hook, build the post, repurpose it, and turn the reach into an owned email list and a sale. The full pack drops Saturday at the $15 founding price. If you want it, comment FOUNDING on the reel and I'll lock you in.
— Conrad · @buildwith.conrad