~/30-days-of-linkedin-posts zeroemdashes.com
Lightning Lesson

30 Days of LinkedIn Posts
in One Afternoon

Everything you need to set up Claude and start generating on-brand posts that actually sound like you.

Bojan Gasparovic  ·  zeroemdashes.com  ·  June 26, 2026

The pre-built Claude OS workspace setup (writing styles, user.md template, post drafting setup): zeroemdashes.com → Zip or Workshop

01
Phase
Set Up Your Claude OS

Claude starts every session cold. It doesn't know who you are, what you work on, or how you want things written. The fix is a context library: a folder of files Claude reads before doing anything else. That's your Claude OS.

What to create
Folder structure
context-library/
  CLAUDE.md           # Claude reads this first, every session
  user.md             # who you are
  writing-style.md     # how you sound
  linkedin-posts/
    posts/             # your drafts and ideas
    linkedin-algorithm.md # the rules (see Phase 4)
  newsletter/
    issues/             # drafts, sent issues
    tone-guide.md       # voice and format rules
  consulting/
    clients/            # one folder per client
    proposal-template.md # reusable starting point
Why this matters
This is the foundation. Every other phase builds on it. Without a context library, Claude produces generic output and you spend the first 10 minutes of every session re-explaining yourself.
Want this set up for you? The 2-hour workshop walks through the full Claude OS setup live. You leave with a working context library, writing styles, and 30 post ideas drafted. From Claude-Curious to Claude-Ready →
02
Phase
Revamp Your LinkedIn Profile

Paste your current LinkedIn profile into Claude: About section, headline, featured. Ask it to rewrite everything. Sharper headline, stronger positioning, a bio that reads like a person wrote it. This is the actual work. The user.md falls out of it automatically - Claude already knows your niche, audience, and voice from the rewrite. Ask it to capture that in a file before you close the session.

What to ask Claude
# In Claude Code, with your context-library open:

1. Paste your LinkedIn profile (About, headline, featured section)
2. Ask: "Rewrite this. Sharper headline, stronger About section,
    better positioning. Then produce a user.md that captures
    my niche, audience, voice, and what I never say."
3. Review both outputs. Update your actual LinkedIn profile.
4. Save the user.md to your context-library root.
What user.md captures
03
Phase
Add Writing Styles

Even with a user.md, Claude's default output often sounds like AI: too polished, too structured, too many insights that land too neatly. A writing style file fixes this by giving Claude a precise model of your sentence rhythm, the things you say, and the things you never say.

What to capture in your writing style file
Pro tip
Read your best-performing posts aloud. The patterns you notice - the things that feel right - are what go in the style file. The file doesn't describe what Claude should do. It describes what you actually do.
The zeroemdashes zip includes ready-to-use writing style files you can load immediately. Available at zeroemdashes.com.
04
Phase
Apply the 360 Brew Rules

LinkedIn replaced its entire feed algorithm in 2025 with 360 Brew: a 150-billion-parameter model that scores each viewer individually. It no longer shows your post to everyone interested in a topic. It asks: is this the right post for this specific person, right now?

Micro-niche wins. Broad loses. And success isn't measured in the first 24 hours - DMs from content typically arrive 3-5 days after posting.

What gets rewarded
Saves - 5x stronger signal than likes
Dwell 61+ seconds - significantly outperforms sub-3s
Substantive comments - 70+ word replies, threads over single comments
Private DM shares - highest-intent signal the algorithm measures
Carousels - 2-3x more dwell than text; 9 slides = 60+ seconds
Native document posts - 5.8% conversion, highest of any format
Long-form 600-1000+ words - ranks higher than short text on frameworks
Topic consistency - 2-3 pillars for 90+ days to build reach beyond followers
Experience-based comments on others' posts - 70+ words, within 20 min
What gets penalised
Links in post body - ~60% reach reduction
Links in comments - also penalised. No safe place for links.
Same structure every time - signals AI content even when it isn't
Engagement bait - "Comment YES if you agree" gets flagged
Generic AI-templated content - deprioritised for lack of personal signal
Off-niche personal stories - doesn't tie to your niche = penalised, not rewarded
Third-party scheduling tools - automation timing patterns detected. Post natively.
Pushing posting volume to recover reach - accelerates the decline
Company page posts - 60-66% lower reach than personal profiles
zeroemdashes.com  ·  Bojan Gasparovic Source: linkedin-algorithm-2026.md
What's next