Writing LinkedIn Posts That Don't Sound Like AI

Where this fits

You've built an agent that drafts weekly LinkedIn updates for a product team. The first few posts were fine, but after a month they all read the same: bullet points, a rhetorical question, and a call to action that screams “engagement bait.” Your users start editing every draft from scratch. That's the problem the linkedin skill from the EveryDay-Writer collection solves. It gives agents a disciplined framework to produce posts that sound like a specific human—your user—rather than a marketing bot.

Why agents benefit

• Enforces a dependency chain before writing. The skill forces the agent to read anti-ai-rules.md, ai_slop_commandments.md, and the user's voice-profile.md first. That means the output isn't just styled—it's constrained against common AI tells like hedging, empty superlatives, and fake urgency.
• Stops fabrication at the source. Section 0.2 of the anti-AI rules forbids inventing stats, quotes, or experience. Agents won't hallucinate authority when the skill's rule book is loaded.
• Prevents over-correction. Rules 9 and 10 in the anti-AI guide tell the agent when not to strip personality or over-polish. The result keeps the writer's genuine tone, not a sterile corporate voice.
• Adapts to direct invocation or dispatch. Whether your agent calls /ew:linkedin or goes through the /ew router, the same constraints apply—consistency across workflows.
• Extensible with reference files. Users can drop .md files in references/ to inject domain-specific examples or past approved posts, giving the agent concrete ground truth to emulate.

Practical scenarios

Career announcement without the cringe A developer is moving to a new company. The agent drafts a post that mentions the role change, thanks the old team, and expresses excitement—without three exclamation marks or a “humbled and honored” cliché. The dependency chain keeps the tone humble but confident.

Thought-leadership piece with an actual opinion Your agent needs to write a 1,500-word article on “why async communication degrades code reviews.” Instead of generic pros and cons, the skill's rules force the agent to take a stance, back it with the writer's experience (from reference files), and avoid bullet-point padding.

Company announcement that doesn't sound like a press release Your startup just closed a funding round. The skill instructs the agent to lead with the customer problem solved, not the money raised. Anti-AI rules kill phrases like “we're thrilled to announce” and force a concrete detail about the product roadmap.

Add it to your agent workflow

Install the skill with:

npx skills add Deupaxx/EveryDay-Writer --skill linkedin

Then in your agent's prompt or system message, call it like this:

Use the linkedin skill to write a post for my profile. Keep to my voice profile and reference my last three posts in the references folder. Draft the post now.

The skill will automatically pull in the dependency chain. For more context on how this fits into the broader skill ecosystem, check out Deupaxx's LinkedIn skill page on OpenAgentSkill.

Compare before adopting

Before you wire this into production, compare it against alternatives. Look at quality signals: does the skill's output reasonably avoid the 12 AI-slop commands? Check maintenance freshness—the repo has 28 stars and is actively updated (v0.3.0). Also consider alternatives like generic LinkedIn prompt templates or other writing skills; they might be lighter but won't enforce the anti-fabrication rules. Finally, assess workflow fit—does your agent have access to the user's voice profile and references? If not, the skill will force an onboarding step that might slow down initial runs.

Why it is worth tracking

The EveryDay-Writer project solves a real pain point: most AI-generated professional content is instantly recognizable and damaging to a personal brand. With 28 stars and a clear versioning cadence, it's early but promising. If you're building agents that write on behalf of humans, track this skill to see how the anti-AI rules evolve—especially the balance between constraint and natural voice. Evaluate it again once the repository crosses 100 stars or adds more skill variants for other platforms.