EW Skill — X / Twitter

Dependency chain. Read these in order before writing anything:

1. core/runtime-contract.md (short operating contract; does not replace the full rules)
2. core/obsidian-workspace.md (resolve storage and persistence rules)
3. core/anti-ai-rules.md (full standard)
4. core/ai_slop_commandments.md (mechanism reference and diagnostic checklist)
5. core/voice-profile.md (voice resolver; require the resolved profile's Completed: field to be exactly Yes, otherwise stop and run onboarding)
6. Any .md files in the active voice's references folder, as resolved by core/voice-profile.md

This chain applies on direct invocation (/ew:tweets) exactly as it does on dispatch through /ew. Direct invocation skips routing, not constraints.

---

WHAT THIS SKILL IS FOR

Standalone tweets and threads that carry actual ideas in minimal space. The discipline here is compression — every word either earns its place or it's out.

X is the fastest feedback loop for writing quality that exists. A weak idea disguised as a smart one dies in minutes. A genuinely sharp observation travels without promotion. There is no algorithm to hide behind.

---

THE UNIT OF CURRENCY: THE IDEA

X rewards one thing above everything else: ideas that feel both new and obviously true once stated. Not motivational. Not clever for the sake of clever. Ideas that make the reader stop and think "I hadn't put it that way but that's exactly right."

Before writing a single word, answer: what is the actual idea? Not the topic — the specific, arguable, non-obvious claim. If you can't state it in one sentence before you write the tweet, you don't have a tweet. You have a vague feeling about a topic.

---

STANDALONE TWEET ARCHITECTURE

The compression test: Can you cut 20% of the words without losing the idea? If yes, it's not tight enough yet. The best tweets have no slack — every word is load-bearing.

The formats that work:

The Inversion: Take conventional advice and flip it. "Stop trying to be consistent. Try being honest. Consistency will follow."

The Specific Observation: Something you noticed that others have seen but not articulated. "The writers who complain most about not having time to write are usually the ones with the most to say. That's not a coincidence."

The Counterargument: Take a widely-held belief and dismantle it with specifics. "'Find your niche' is advice that made sense when there were fewer writers. Now the niche finds you if the writing is good enough."

The Compression of a Complex Idea: Collapse something nuanced into the fewest words that don't betray it. "Most first drafts aren't bad writing. They're correct writing about the wrong thing."

What doesn't work:

• Inspirational statements with no claim ("Dream big. Work hard. The rest follows.")
• Hot takes without logic ("[Thing everyone likes] is actually terrible")
• Thread bait that buries the idea ("A thread on why most writers fail: 🧵")
• Anything that starts with "Unpopular opinion:" — the opinion had better be actually unpopular
• Lists in standalone tweets — compress the list into a sentence

---

THREAD ARCHITECTURE

A thread is a long-form argument broken into beats, not a list padded out with line breaks.

Thread rules:

Tweet 1 is everything. The entire thread must earn its existence in the first tweet. State the full claim — not a tease, the actual point. "Here's why X happens" is a tease. "X happens because Y, and most people have the fix backwards" is a claim. Readers decide in the first tweet whether to continue.

Each tweet advances the argument. Not restates it. Not transitions to it. Advances it. If tweet 4 says roughly the same thing as tweet 2 in different words, cut one.

Stair-step the stakes. Each tweet should feel more specific, more surprising, or more actionable than the one before it. The payoff at the end must be the sharpest thing in the thread — not a summary, not a recap, the hardest-earned point.

No tweet is a filler. "This is why it matters" is not a tweet — it's a transition thought that belongs inside another tweet. Cut it. Every tweet stands alone and earns its place in the sequence.

Last tweet: The sharpest single insight from the whole thread, standalone. Someone who only reads the last tweet should feel like they got something.

Ideal thread length: As short as it can be while fully developing the idea. 5–8 tweets for most arguments. A 15-tweet thread usually contains a 5-tweet thread and 10 tweets of self-indulgence.

---

COMPRESSION MECHANICS

Kill the setup. Most tweets spend 60% of their character count getting to the idea. Cut the setup and start with the idea. ❌ "I've been thinking a lot about productivity lately, and I've come to realize something important:" ✅ "Most productivity advice is designed for people who aren't actually avoiding the work."

One idea per tweet. Two ideas in one tweet means two tweets that are each weaker. Split them.

The word "just" test. Read every tweet and ask: what does "just" do here? Almost always, nothing. Delete it. Same test applies to "really," "very," "quite," "actually," "basically."

The active construction test. Every passive construction is an opportunity to make the sentence more direct. "Mistakes were made by the team" → "The team got it wrong." In 280 characters, passive voice is a waste of space.

---

VOICE ON X

X rewards a distinct voice more than any other platform. The writers with the largest audiences on X are not the most informed — they're the most distinctly themselves. Your voice is your competitive advantage. Don't sand it down to fit a template.

What creates a distinct voice on X:

• A consistent perspective — a set of beliefs the reader can anticipate and trust
• Specific details over general claims — named examples beat unnamed ones every time
• Earned confidence — opinions stated as opinions, not dressed up as facts
• Humor that comes from insight, not performance

What destroys a distinct voice on X:

• Chasing trending topics without a genuine angle
• Responding to everything — dilutes the signal
• Thread formats copied from successful accounts — the format isn't the reason it works
• Posting for engagement metrics — readers can feel the difference

---

POST-GENERATION REVIEW

After producing any draft, stop. Re-read core/anti-ai-rules.md Sections 1–7 with the draft in front of you. Go line by line through the pre-publish checklist in Section 7. Fix every failure before presenting the output.

On X, a single AI-patterned phrase kills the post's credibility instantly. The review pass catches what the generation pass doesn't.

PRE-POST CHECKLIST

• What is the single specific idea this tweet or thread makes? State it before writing.
• Is every word load-bearing — does cutting anything lose the meaning?
• For threads: does tweet 1 state the full claim, not just tease it?
• For threads: does each tweet advance the argument, not restate it?
• Does the last tweet/thread tweet contain the sharpest point?
• Have you run Sections 1–7 of core/anti-ai-rules.md?
• Are there any filler words ("just," "really," "actually") that can go?
• Is this trying to be clever, or does it have a real idea?