When NOT to use

• You need content strategy or pillar planning — use social-content-os
• You are writing long-form blog posts or documentation — this is social-native copy
• You need video scripts — see video-specific skills
• You are drafting email newsletters — use newsletter-craft
• The post has no connection to real work, opinions, or proof

Core concepts

The hook

• The first line is the entire post for the majority of readers. It must earn the click to expand.
• Hook patterns that work for builders: Contrarian claim; Metric lead; Before/after; Direct how-to; Question hook; Story opener.
• Anti-patterns: "I'm excited to announce…"; "Thread 🧵"; filler openers ("So…", "Alright…"); humble-brag framing.

Platform signals update and what to change in practice

• LinkedIn: prioritize dwell time and genuine discussion. Longer narrative, carousels (PDF posts) and comment-driven discussion increase dwell. Avoid obvious engagement-bait ask-for-comments posts — the new ranking deprioritizes them (see ALM Corp analysis).
• X: prioritize early engagement in the first 30–60 minutes and craft posts that invite quick reactions. Account subscription/verification status is an important signal; pay-centric distribution means organic reach is smaller for non-subscribers (see Sprout Social).

X drafting rules (updated)

• Single post (short-form): keep it tight. Many readers still prefer compact posts; avoid filler. Note: X supports longer posts, but shorter first-line clarity wins attention.
• Early-engagement rule: the first 30–60 minutes of reactions (likes, replies, reposts) strongly influence distribution. Arrange a small, genuine cohort to engage early if appropriate.
• Thread (3–12 posts):
  1. Post 1 (hook): standalone insight or claim
  2. Posts 2–N: one idea per post, each readable in isolation
  3. Final post: actionable takeaway or CTA
  4. Line breaks between sentences; readers scan
  5. Use numbering only when order matters
• Thread length caution: reader completion drops fast after ~6–8 posts. If you need more, split into a series and label it.
• Hashtags: use sparingly and only when they are semantically relevant to discovery.

Thread pacing template (unchanged) Post 1: Hook — the bold claim or result Post 2: Context — the problem Post 3: What we tried first (why it failed) Post 4: The actual solution (with specifics) Post 5: Result — metrics, before/after, proof Post 6: Takeaway — the lesson Post 7: CTA — what the reader should do next

LinkedIn drafting rules (updated)

• Format: lead with a strong hook line, then a line break (LinkedIn truncates early). Short paragraphs, aggressive white space.
• Tone: more narrative than X. Vulnerability works; show a learning arc.
• Proof and dwell: include metric + narrative + an embedded artifact (screenshot, carousel) to increase dwell time and comment depth.
• Engagement: end with a question or explicit, contextual CTA. Avoid low-value ask-for-comments prompts; the ranking now favors authentic discussion.
• Link handling: LinkedIn deprioritizes external links in the body. Put external links in the first comment, or include a clear CTA to follow the profile for the link.

LinkedIn-specific formats that perform (updated)

• Listicle: "7 things I learned building [X]" + numbered list
• Carousel (PDF): 6–12 slides, 1 idea per slide — increases dwell and is favored under the new ranking
• Story arc: Setup → Conflict → Resolution → Takeaway
• Hot take + context: Bold claim → reasoning → "Agree or disagree?"

Tone calibration

• X: dense, compressed, technical where appropriate, minimal self-reference, witty/terse humor
• LinkedIn: expanded, narrative, vulnerable, professional but human
• CTAs: X: "Bookmark this" / "Follow for more"; LinkedIn: "What's your experience?" / "Drop your take"

Proof-backed writing

• Always attach proof: screenshot, metric with baseline, repo/link, or a user quote. If no proof exists, label the post as opinion.

Workflow Step 1 — Pull from backlog: ensure hook, pillar, platform, proof Step 2 — Write 3 hook variants; pick the one with the highest specificity-to-curiosity ratio Step 3 — Draft body per platform rules Step 4 — Tone check: read aloud Step 5 — Platform adapt: write X first (constraint), expand for LinkedIn Step 6 — Proof attach: screenshot for X; carousel or metric narrative for LinkedIn Step 7 — Schedule:

• X: post when followers are active; prioritize morning or lunch local time and be ready to respond in the first hour
• LinkedIn: Tue/Wed/Thu, 8–10 AM local. Carousels and long narratives perform best on weekdays

Examples (updated with signal-aware notes) Example X thread (5 posts) Post 1: I replaced our entire RAG pipeline with 50 lines of code. It's faster, cheaper, and more accurate. Here's exactly how:

(Note: prepare to amplify early engagement in the first hour — distribution depends heavily on early velocity.)

Example LinkedIn post I spent 6 months building a custom RAG pipeline. Last week, I deleted most of it. Here's why that's a good thing: (Include carousel or screenshot to increase dwell time.)

Decision tree

1. Do I have a scored backlog item? No → Content OS. Yes → continue.
2. Platform: X, LinkedIn, or both? Pick one to draft first.
3. Proof artifact present? No → create one or switch to opinion framing. Yes → continue.
4. Single point or multi-step story? Single → X single post or LinkedIn short. Multi → X thread or LinkedIn long.
5. Adapting for both? Yes → write X first, expand for LinkedIn.
6. Hook pass the "would I click expand?" test? No → rewrite. Yes → ship.

Edge cases and gotchas (updated)

• Thread fatigue: X threads longer than 8–10 posts lose readers sharply; consider serializing.
• LinkedIn link penalty: avoid external links in body; put links in first comment.
• Self-promotion ratio: keep ~4:1 value to promotional posts.
• Cross-post detection: adapt copy — audiences notice copy-paste.
• Image alt text: add manually on X for accessibility and search signals.
• Quote-tweet trap: add your take when quote-tweeting; short reactions add little value.
• LinkedIn LLM sensitivity: obvious engagement-bait and recycled, low-value content are downranked (see ALM Corp analysis).
• X pay-centric model: non-subscribers have lower baseline organic reach; early engagement and subscription status matter (see Sprout Social).
• Show HN: Use "Show HN: <title>" and make your project easy to try without barriers (see HN Show guidelines).
• Product Hunt: follow the Product Hunt Launch Guide — prepare supporters and a day-of plan; use the official checklist for launch-day mechanics.

Evaluation criteria (updated)

• Hook click-through (expand rate): Poor <5% | Acceptable 8–15% | Strong >20%
• X thread completion rate: Poor <30% | Acceptable 40–60% | Strong >70%
• LinkedIn comment count: Poor 0–2 | Acceptable 5–15 | Strong >20
• Proof artifact present: No | Screenshot only | Metric + screenshot + link
• Platform adaptation: Copy-paste | Minor tweaks | Fully rewritten per platform
• Tone accuracy: Sounds like marketing | Sounds professional | Sounds like a builder talking to peers
• CTA effectiveness: No CTA | Generic "follow me" | Specific, contextual CTA

Quick reference (copy-paste templates)

• X single: "Metric lead + one-sentence takeaway + CTA" (e.g., "Cut API latency 2.3s → 180ms. The fix was simple: delete the reranker. Bookmark for the code.")
• X thread opener: short bold claim that stands alone
• LinkedIn opener: one-sentence hook then line break; follow with short paragraphs and a carousel if available

Sources and references

• ALM Corp: LinkedIn Feed Algorithm Now Uses LLMs (Mar 14, 2026): https://almcorp.com/blog/linkedin-feed-algorithm-update-llm-2026/
• Sprout Social: How the Twitter (X) Algorithm Works in 2026: https://sproutsocial.com/insights/twitter-algorithm/
• Hacker News: Show HN Guidelines: https://news.ycombinator.com/showhn.html
• Product Hunt Launch Guide (official): https://www.producthunt.com/launch