twitter-thread-style
Skill57 starsChanged 5 months agoMITView on GitHub
Write or revise X/Twitter developer announcement threads in the user's preferred style. Use for technical launch threads, project announcements, release threads, and threads with code screenshots/snippets.
In this repository15 other files it gives its agents.
- AGENTS.mdAGENTS.md
- agent-browserSkill
- ai-newsSkill
- applescriptSkill
- birdSkill
- chat-to-skillSkill
- dont-anger-marioSkill
- github-issuesSkill
- keyboard-layout-decoderSkill
- maintainer-replySkill
- package-releaseSkill
- pr-reviewSkill
- session-reflectSkill
- skill-discoverySkill
- vibe-mergeSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
15 rules. About 0 are common, 15 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈15
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 8
Use this skill to write developer-facing X/Twitter threads in the user's voice.
- Distinctive: line 14
Write a technical announcement thread that works in the timeline.
- Distinctive: line 61
Do not inline code in prose.
- Distinctive: line 138
Use aliases/imports for readability when the language supports them.
- Distinctive: line 160
Avoid snippets that only show plumbing unless plumbing is the point.
- Distinctive: line 164
Keep prose short. Let code carry meaning.
- Distinctive: line 171
If the prose explains something the code already shows, delete the prose.
- Distinctive: line 175
Do not lead with architecture unless the architecture is the product.
- Distinctive: line 208
Use comparisons to explain, not to posture.
- Distinctive: line 230
Do not explain a project with abstract positioning first.
- Distinctive: line 259
Do not force every project into a personal-pain story.
- Distinctive: line 275
Do not put limitations in the first tweet unless the limitation is central to understanding the project.
- Distinctive: line 281
Avoid pretending a roadmap item is implemented.
- Distinctive: line 297
If the user pushes back, treat it as style data.
- Distinctive: line 309
Do not argue. Rewrite.
≈ “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
Twitter Thread Style
Use this skill to write developer-facing X/Twitter threads in the user's voice.
This is a generic style skill. Do not bake in one project's facts. Read the current task context, source material, or user notes first, then apply the style rules below.
Goal
Write a technical announcement thread that works in the timeline.
The reader should be able to understand the project from the first tweet, but the code snippets/screenshots should be what make them stop scrolling.
The thread should feel like a developer showing a real tool, not a company writing a launch page.
Voice
Prefer:
- plain technical statements
- concrete reasons the thing exists
- direct comparison to a known tool/category when it helps
- understated confidence
- real tradeoffs
- specific capabilities
- attractive code snippets
Avoid:
- slogan speak
- marketing copy
- generic hype
- vague architecture-first explanations
- buzzwords like “unlock”, “seamless”, “delightful”, “empower”, “revolutionary”
- claims that sound bigger than the evidence
- explaining internals before explaining why the reader should care
Bad tone:
This unlocks a delightful new developer experience for modern teams.Better tone:
I wanted the same workflow, but without leaving the stack I already use.Strict Tweet Shape
Each tweet must follow this shape:
- A short chunk of text.
- Then one or more code blocks.
Do not inline code in prose. No backticked identifiers inside the paragraph.
Bad:
Run `my_tool dev` and edit `pages/index.md`.Good:
The dev server is one command.my_tool devA tweet can contain multiple code blocks. Use this to increase density.
First Tweet
The first tweet must be optimized for the timeline.
It should quickly answer:
- what was released
- what familiar category/tool it relates to
- what makes it different
- what the code feels like
The first sentence should be short and direct.
Good patterns:
I published PROJECT.PROJECT is public.FAMILY has a new member: PROJECT.Then use a familiar anchor if possible:
It is like KNOWN_TOOL, but for STACK.I wanted KNOWN_WORKFLOW, but with DIFFERENT_CONSTRAINT.The hook should usually be capability-first, not positioning-first. Lead with what became possible, what shipped, what changed, or what the thing is in familiar terms.
Then show the best code snippet immediately.
Code Is the Hook
Prioritize snippets that look good as screenshots.
Good snippets usually show:
- compact syntax
- normal language constructs instead of wrapper DSL noise
- typed schemas or validation
- route generation
- small configuration blocks
- pipelines with readable aliases/imports
- components/templates with attractive syntax
- before/after transformation
- one command install/build/dev flow
Use aliases/imports for readability when the language supports them.
Bad:
value =
input
|> Very.Long.Namespace.Module.step_one()
|> Very.Long.Namespace.Module.step_two()Better:
alias Very.Long.Namespace.Module
value =
input
|> Module.step_one()
|> Module.step_two()Avoid snippets that only show plumbing unless plumbing is the point.
Text Density
Keep prose short. Let code carry meaning.
A tweet should usually be:
- 1 to 4 short prose lines
- 1 to 3 code blocks
If the prose explains something the code already shows, delete the prose.
Architecture Comes Later
Do not lead with architecture unless the architecture is the product.
Bad early framing:
The split is intentional.The system has three layers.Better early framing:
The default workflow stays inside one toolchain.The config is just normal language code.Explain architecture through user-visible consequences:
- fewer processes
- fewer config files
- clearer data flow
- better integration with existing tools
- less generated glue
- safer boundaries
Comparisons
Use comparisons to explain, not to posture.
Good:
I wanted the KNOWN_TOOL workflow, but in STACK.It sits in the same space as KNOWN_TOOL, but the API follows STACK conventions.Avoid:
It is better than KNOWN_TOOL.unless the next line proves a very specific point.
Concrete Driver
Do not explain a project with abstract positioning first. Explain the concrete driver.
Good drivers include:
- a new capability
- a familiar workflow moved into a different stack
- a measurable improvement
- a missing abstraction extracted from real work
- a new release or integration
- a personal itch or frustration
Examples of good driver shapes:
PROJECT can now DO_THING.I wanted KNOWN_WORKFLOW, but in STACK.I got roughly N× improvement by changing THING.PROJECT now supports INTEGRATION.Do not force every project into a personal-pain story. Pick the real driver for this announcement.
Story Arc
A good technical announcement thread usually follows this arc:
- Hook: project/category/difference plus best code.
- Concrete driver: what became possible, simpler, faster, extracted, or newly supported.
- Install or minimal usage.
- Main syntax/API.
- Data/configuration model.
- One or two advanced capabilities.
- Integration with the surrounding ecosystem.
- Honest limitations or current scope.
- Links, version, repository/package.
Do not put limitations in the first tweet unless the limitation is central to understanding the project.
Honesty
Be precise about what exists now and what is future work.
Avoid pretending a roadmap item is implemented.
Good:
This part is static for now. Runtime support needs a separate design.Good:
The first version is small, but the core path works.When Revising
If the user pushes back, treat it as style data.
Common meanings:
- “Why should the end user care?” means the draft is too architecture-first.
- “Too much slogan speak” means replace marketing phrases with concrete technical facts.
- “No inline code” means prose must not contain backticked identifiers.
- “Use alias for readability” means snippets should be screenshot-readable.
- “Prioritize impressive code snippets” means move the best code earlier and shorten the explanation.
- “Too specific” means abstract the rule; do not force one announcement pattern onto every thread.
- “Generic skill” means remove project-specific facts and keep reusable style rules only.
Do not argue. Rewrite.
Output Format
Use numbered tweets.
Template:
1/
Short prose. No inline code.
```language
code block
```
```language
optional second code block
```
2/
Short prose.
```language
code block
```Before finalizing, check:
- Does the first tweet have a strong code snippet?
- Is every tweet text-first then code?
- Is there no inline code in prose?
- Is the prose concrete rather than slogan-like?
- Are code snippets readable as screenshots?
- Did the thread explain why the reader should care before internals?
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.

