agentleFS
jsathjsath/claude-skills/skills/writing/tweet-writer/SKILL.md

tweet-writer

Skill14 starsChanged 34 days agoMITView on GitHub

Writes X (Twitter) posts and threads in the user's own voice, learned from 5 to 10 of their best posts, using format rules measured on real post data: numbered tutorials and copy-paste config tips beat stories, numbered lists beat arrow lists, link cards die. Use when someone asks for a tweet, a thread, tweet ideas, or to turn a demo, lesson or result into an X post. Drafts only; never posts.

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

7 rules. 0 are common, 5 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Ask the user to paste 5 to 10 of their own top posts (highest views or bookmarks.

    line 14
  2. Distinctive:

    Bookmarks are the metric: Write things people save: exact steps, a config, a checklist.

    line 43
  3. Distinctive:

    Never: em dashes, "delve", "leverage", "utilize", "it's worth noting", "thread 🧵", rocket or lightbulb emojis, more than one hashtag.

    line 55
  4. Distinctive:

    No grand lesson, no slogan stack at the end.

    line 100
  5. Distinctive:

    No numbers: write it without them.

    line 114
  6. Distinctive:

    No voice samples: use the defaults, label the draft "default voice".

    line 115
  7. Very specific:

    When the user rejects or edits a draft, or shares how a post performed, append a line to feedback.md (date, post format, result or …

    line 120

“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

Writes posts that sound like you and use the formats that actually get read and saved on X.

By Johann (@johannships). The format rules come from pulling and ranking 100 of my own posts by views, likes and bookmarks, then re-checking against newer posts. They held across months.

Step 0: learn the voice (first run, then reuse)

Ask the user to paste 5 to 10 of their own top posts (highest views or bookmarks; X Analytics or the post itself shows them). Then write a short voice.md next to this file with:

  • capitalisation habit (all lowercase? sentence case?)
  • typical line length and whitespace pattern
  • words and shorthand they use (e.g. "&", "gonna", "tbh") and words they never use
  • emoji and hashtag habit (count them, do not guess)
  • how they open and how they close

Show voice.md to the user once and fix anything they correct. On later runs, read it instead of asking again. If they will not share posts, use the default voice at the bottom and say so.

Inputs

  • Topic, result, demo or lesson (required). Real numbers if they have them.
  • Format: single (default), thread, or ideas.
  • Optional: their own draft, hook or CTA. Anything they supply is locked: keep their hook word for word, keep their bullet structure and CTA. Shorten inside those parts before cutting any of them.

Formats ranked by measured performance

RankFormatWhat it looked like in the data
1Numbered step-by-step tutorial with an insider hook ("the full X setup nobody explains:")Every top post in the sample was one of these two formats
1Copy-paste config tip + before/after ("X tip most people miss:" then the exact snippet)Bookmarks ran about 2x likes. Re-posting variations kept working; the format did not fatigue
2Numbered process ("i automated my X:" then 1 to 10)Solid
3Contrarian "stop doing X" + the fixSolid
4Bullet list of use casesDecent
5Arrow (->) listAbout 4x worse than the same content as a numbered list
6Pure story with no structureWeak
7Link cards, product announcements, landing-page screenshotsDead last every time

Rules that follow from it:

  • Bookmarks are the metric. Write things people save: exact steps, a config, a checklist.
  • Always numbered lists, never arrows.
  • A story works only with structure: a real scenario, then a numbered list of what happened, then a one-line reframe.
  • Never lead with the product. Show the thing working, not the landing page.
  • Fear or money hooks work when the list delivers ("your X setup can be hacked in 5 minutes. 10 things to lock it down:").

Voice defaults (overridden by voice.md)

  • Whitespace is mandatory: one thought per block, blank line between blocks, blank line between numbered items.
  • Real numbers over vague ones ("208 hours in calls", not "lots of calls"). Only numbers the user gave you.
  • Get to the point. No "here's what i learned about X" setup.
  • Observations over advice ("i noticed X" beats "you should do X").
  • Never: em dashes, "delve", "leverage", "utilize", "it's worth noting", "thread 🧵", rocket or lightbulb emojis, more than one hashtag.

Correct:

unpopular opinion: the best software companies will be run by 1 person

with agents doing everything else

we're already here

Wrong (same words, no air):

unpopular opinion: the best software companies will be run by 1 person with agents doing everything else. we're already here.

Shapes

Tutorial:

the full [thing] setup nobody explains:

1. [step]

2. [step]

3. [step]

[one-line payoff]

Demo caption ("i built..."):

i built [thing] that [does job]

you tell it "[plain-English instruction]"

then it:
- [mechanism 1]
- [mechanism 2]
- [mechanism 3]

[one real use line]

[simple reply CTA]

No grand lesson, no slogan stack at the end.

Thread: tweet 1 is the hook (never "a thread on..."), one point per tweet, last tweet is the payoff or a soft CTA. Each tweet under 270 characters.

Output rules

  • Return only the post text, ready to paste. No labels or rationale unless asked.
  • "give me the tweet" or "just the tweet" = exactly one post.
  • Ideas mode = 5 to 10 options, each a different format or angle, not rewrites of one thesis.
  • Count characters mechanically (whitespace, bullets and emoji included). Single post: under 280 unless the user has X Premium and asks for long-form. Never claim a long post fits 280.
  • Long-form asked for ("a bit more detail", "like I'm talking"): keep experience, what changed, takeaway in spoken sentences with spacing. Do not compress the story into slogans to hit 280.

When something is missing

  • No numbers: write it without them. Never invent views, revenue, users or time saved.
  • No voice samples: use the defaults, label the draft "default voice".
  • User says it sounds like AI: simplify. Fewer staged one-liners, fewer contrasts, plainer words. Do not add fake casualness ("lol").

Feedback log

When the user rejects or edits a draft, or shares how a post performed, append a line to feedback.md (date, post format, result or correction, rule going forward). Read it before writing.

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.