x-hook
Skill1 starsChanged 6 months agoMITView on GitHub
Write or rewrite the first line (hook) of an X (Twitter) post. Use when the user has content, an idea, or a finished post and wants a stronger opening, asks "make this hook better", "how should I start this tweet", "my posts get no engagement", or wants scroll-stopping first lines. Returns ranked hook variants with reasoning.
In this repository5 other files it gives its agents.
- x-auditSkill
- x-profileSkill
- x-replySkill
- x-teardownSkill
- x-threadSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
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
- Distinctive: line 24
Use when there is real data.
- Distinctive: line 41
No throat-clearing: cut "I think", "Just wanted to share", "Sorry for the long post".
- Distinctive: line 43
Don't bait. "Like if you agree" / "comment X to get it" trains negative actions.
- Distinctive: line 61
If the user gave you a finished post, also show how the rest of the post should adjust to pay off the recommended hook.
≈ “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.
Files
x-hook — Hook Engineering
The first line is the highest-leverage part of any X post. Read reference/ALGORITHM.md
if available. The mechanism: Phoenix ranks per viewer, and a viewer decides to stop,
expand, or swipe within the first line. A swipe-away trains not_interested; an
expand-and-read trains dwell and click. The hook decides which one fires.
What you need from the user
The content or idea the post is about, and ideally: the niche, the audience, and any concrete facts/numbers/names involved. Ask if missing — you cannot write a specific hook from a vague brief.
The 7 hook archetypes
Pick the 2–3 that fit the user's material and write variants in each.
- Specific number + outcome —
I [verb] [exact number] in [timeframe] doing [thing].Use when there is real data. Highest entity density → sharp embedding. - Counter-consensus —
Everyone says [X]. They're wrong.Use only if the user can defend the claim when quote-tweeted. - Breakdown promise —
I analyzed [N] [things]. Here are the [M] patterns:Strong save/bookmark andshare_via_dmdriver. - Vulnerable confession —
I almost [failed] last [period]. Here's what I learned.Real failure only; never manufactured. - Screenshot + one line — A caption under a screenshot. The default workhorse format.
- Concrete before/after —
[Old state] → [new state].Visual, compressed. - Named contrast —
[Tool A] vs [Tool B], after [N] [hours/projects]:Entity-dense, retrieves into both tools' audiences.
Rules for every hook
- One concrete entity minimum in line 1 — a number, name, tool, or date. This sharpens the retrieval embedding and is the difference between reaching the right cluster and no one.
- No
🧵, no "thread below", no "a quick thread on". In 2026 these read as "long ad, skip". - No throat-clearing: cut "I think", "Just wanted to share", "Sorry for the long post".
- It must stand alone. Assume the viewer sees ONLY line 1.
- Don't bait. "Like if you agree" / "comment X to get it" trains negative actions.
- Honest. A hook that overpromises gets a swipe-away on line 2, which is worse than a modest hook.
Output format
Material read: <one line on what you understood>
Hooks (best first):
1. "<hook>"
Archetype: <name> · Why it works: <one line, mechanism-grounded>
2. ...
(5–8 variants)
Recommended: #<n> — <one line on why it fits this user/audience best>If the user gave you a finished post, also show how the rest of the post should adjust to pay off the recommended hook. Never invent numbers the user didn't supply — if a number- based hook would be strongest, say so and ask for the number.
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
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.

