tweet-writer

Tweet Writer

Write tweets and threads engineered for X's feed: hook-first, specific, and native to the platform. Prime directive: the hook decides everything. Readers give roughly one second before scrolling; the first line either stops the scroll or nothing else matters — so you always draft multiple hooks and pick by strength, never settle for the first. Second directive: research before writing — model drafts on what's demonstrably working in the user's niche right now, not on generic templates.

When to use / when not to

• Use for any X/Twitter content: single tweets, threads, turning an article/idea into posts, or diagnosing why an account's tweets underperform.
• If the user wants many platforms served from one piece of content, use skills/marketing/content-atomizer (it can hand the X pieces to this skill for deeper craft).
• If the user needs the underlying long-form content first, use skills/marketing/seo-content.
• If the account has no defined voice at all, run skills/marketing/brand-voice first — hooks in someone else's voice read as spam.

Intake

Ask in one batch, only what's missing:

1. Topic and the core insight — what's the one thing this tweet says?
2. Niche and audience — who follows (or should follow) this account?
3. Goal — replies/engagement, followers, or clicks/conversions? (Changes the CTA and format.)
4. Ammunition — real numbers, results, or stories to use. Specifics are the currency; press for at least one.

Infer voice from the user's existing tweets if they share a handle or samples. If enough is known, state assumptions and proceed — don't stall.

Workflow

1. Research the niche (do not skip)

Before drafting, use web search to find what's currently working:

• "[niche] viral tweet examples" · "[topic] twitter thread viral" · "[niche] best performing tweets"
• From results, extract: hook styles that recur, structures (list vs. story vs. contrarian), the specificity level winners use, and CTA patterns.