agentleFS
sanky369sanky369/vibe-building-skills/skills/marketing/tweet-writer/SKILL.md

tweet-writer

Skill30 starsChanged 5 months agoMITView on GitHub

Write high-engagement tweets and X/Twitter threads, grounded in live research of what's currently working in the user's niche: researches viral examples via web search, then drafts hook variants, a single-tweet version, and a thread version with alternates to test. Use when the user says 'write a tweet about X', 'turn this into a thread', 'help me go viral', 'grow my X/Twitter account', 'my tweets get no engagement', or shares an idea/article to tweet — any X/Twitter content request. Produces a Tweet Package: 3 hook options, final single tweet, full thread, and posting notes.

8 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use TubeScribe, deep-research-pro and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • AlekseiULSkill

    TubeScribe

    YouTube video summarizer with speaker detection, formatted documents, and audio output. Works out of the box with macOS built-in TTS. Optional recommended tools (pandoc, ffmpeg, mlx-audio) enhance quality. Requires internet for YouTube access. No paid APIs or subscriptions. Use when user sends a YouTube URL or asks to summarize/transcribe a YouTube video.

    Also in 3 projects that use this

  • GPTtangSkill

    deep-research-pro

    Multi-source deep research agent. Searches the web, synthesizes findings, and delivers cited reports. No API keys required.

    Also in 3 projects that use this

  • vijpatel7Skill

    audit-website

    Audit websites for SEO, performance, security, technical, content, and 15 other issue cateories with 150+ rules using the squirrelscan CLI. Returns LLM-optimized reports with health scores, broken links, meta tag analysis, and actionable recommendations. Use to discover and asses website or webapp issues and health.

    Also in 3 projects that use this

  • elliottminnsSkill

    gog

    Google Workspace CLI for Gmail, Calendar, Drive, Contacts, Sheets, and Docs.

    Also in 4 projects that use this

  • ReScienceLabSkill

    reddit

    Search and retrieve content from Reddit. Get posts, comments, subreddit info, and user profiles via the public JSON API. Use when user mentions Reddit, a subreddit, or r/ links.

    Also in 3 projects that use this

  • aisa-groupSkill

    presentation

    Presentation creation, editing, and analysis. When Claude needs to work with presentations (.pptx files) for: (1) Creating new presentations, (2) Modifying or editing content, (3) Working with layouts, (4) Adding comments or speaker notes, or any other presentation tasks.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use for any X/Twitter content: single tweets, threads, turning an article/idea into posts, or diagnosing why an account's tweets …

    line 12
  2. Very specific:

    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 …

    line 13
  3. Very specific:

    If the user needs the underlying long-form content first, use skills/marketing/seo-content.

    line 14
  4. Very specific:

    If the account has no defined voice at all, run skills/marketing/brand-voice first — hooks in someone else's voice read as spam.

    line 15
  5. Distinctive:

    Ask in one batch, only what's missing:

    line 19
  6. Distinctive:

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

    line 32

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “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

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.

Summarize findings in 3–5 bullets — this brief steers every drafting choice. If web search is unavailable, say so and fall back to the hook taxonomy in references/tweet-hooks.md, labeling the approach as template-based rather than researched.

2. Choose format by goal and material

If…Format
One sharp insight, goal = engagementSingle tweet (listicle, contrarian take, before/after, or fill-in-the-blank)
Multi-step framework, story arc, or article sourceThread (5–10 tweets)
Goal = replies specificallyQuestion or fill-in-the-blank single tweet
Goal = clicks/conversionsThread with CTA closer; link in the reply, never the main tweet

3. Draft 3 hooks (always)

Generate three genuinely different hooks — different types, not rewordings. Pull from these families (full taxonomy with examples: references/tweet-hooks.md):

  • Specific result: "I [result] in [timeframe]. Here's how:" — strongest when real numbers exist
  • Contrarian/pattern interrupt: "Everyone says [X]. They're wrong." / "Stop doing [common practice]."
  • Curiosity gap: "The one thing [group] gets wrong about [topic]:"
  • Story: "[Time] ago I was [bad state]. Today [good state]."
  • List promise: "[N] [things] that [benefit]:"
  • Extreme: "The best/worst/fastest [X] I've ever [Y]" — superlative, then overdeliver

Hook rules: lead with the payoff, not context; concrete number > adjective ("12,847 followers in 63 days" beats "grew fast"); first 7 words must carry the whole promise. Recommend one hook with a one-line reason; the user picks.

4. Write the body

Single tweet — aim ≤110 characters when the idea allows (heuristic: shorter scans faster and quotes better); apply a frame: PAS (problem → agitate → solution), BAB (before → after → bridge), or bare list. One idea only.

Thread — 5–10 tweets:

  1. Hook tweet — the chosen hook + thread signal ("🧵" or a colon ending)
  2. Context tweet — why this matters, deepen the curiosity
  3. Body tweets (one idea each) — ≤250 chars per tweet; number them (2/, 3/ …); every tweet must earn the next click — end tweets on mini-cliffhangers or completed punches, never mid-mush
  4. Summary tweet — compress the takeaways
  5. CTA closer — exactly one ask: a question (drives replies), "follow for more [specific topic]", or a link — and links go in a reply under the thread, not in tweet 1

Persuasion levers to apply throughout: specificity (real numbers), social proof (real counts/names), curiosity gaps, relatability ("when you finally…"), tasteful contrarianism. All numbers and results must come from the user or the research — never fabricate metrics, follower counts, or outcomes.

5. Package and deliver

Run the quality bar, then deliver the Tweet Package. Include posting notes: platform-mechanics heuristics, clearly labeled as such — early engagement matters (be available to reply in the first hour); replies and bookmarks appear to outweigh likes; external links in the main tweet correlate with suppressed reach (put them in a reply); asking for likes/RTs reads as spam and underperforms. Don't promise algorithmic outcomes — these are patterns, not guarantees.

Required output format

# Tweet Package — [topic]

## Research brief
- [3–5 bullets: patterns found in the niche, or "web search unavailable — template-based"]

## Hook options
1. [Type] "…" (n chars)
2. [Type] "…" (n chars)
3. [Type] "…" (n chars)
Recommended: #_ — [one-line reason]

## Single tweet
"…" (n chars)

## Thread
1/ "…" (n chars)
2/ "…" (n chars)
...
[Link-in-reply text, if goal = clicks: "…"]

## Alternates to test
- [1–2 alternate hook tweets or a variant angle]

## Posting notes
- CTA: [the one ask] · Link placement: [none | in reply]
- [Heuristics that apply: reply-window availability, no engagement-begging, etc.]

Quality bar

Count, don't eyeball:

  • 3 hook options of genuinely different types, each with char count
  • Single tweet ≤280 chars (target ≤110); every thread tweet ≤250 chars
  • First 7 words of the chosen hook carry the promise — read them alone and check
  • At least one concrete specific (number, timeframe, named thing) in the hook or body — or explicitly flagged that the user provided none
  • All numbers/results real (from user or cited research); nothing invented
  • Exactly one CTA, in the closer only; no "like and RT" begging anywhere
  • No external link in the main tweet — link text staged for a reply
  • One idea per tweet; no filler tweets ("let me explain…" alone is not a tweet)
  • Voice matches the user's samples — a follower should believe they wrote it

Integration

  • skills/marketing/brand-voice — supplies the voice; run first for new/undefined accounts.
  • skills/marketing/content-atomizer — feeds this skill article-derived ideas for X; sends back the finished tweets into its distribution pack.
  • skills/marketing/seo-content — published articles are prime thread source material; the thread's link-in-reply points back to the article.
  • skills/marketing/direct-response-copy — deepens conversion-focused CTAs when a tweet/thread sells directly.

References

  • references/tweet-hooks.md — the full 8-framework hook taxonomy (Extreme, AIDA, PAS, BAB, and more) with structures and examples. Read it at Step 3 when drafting hooks, or whenever the three drafts feel same-y and you need a different attack angle.

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.