linkedin_publishing v1.0.0

name: linkedin_publishing description: Publish on LinkedIn that actually travels. Drafts high-performing posts (hooks, format picks backed by 2026 performance data), writes comments that grow reach, and runs company page ops (admin roles, invites, employee advocacy). Triggers when someone asks for LinkedIn content help: write a post, plan a week of content, improve engagement, manage a company page, or comment strategically.

LinkedIn Publishing

Purpose

Turn a topic into LinkedIn content with real reach, or run the operations behind a company page. Covers three jobs: personal-profile posting, comment strategy, and company page ops. The approach is data-driven, not vibes: format choices come from 2026 performance studies (Metricool, Socialinsider, AuthoredUp), and every draft passes through the bin/check_draft.py linter before it ships.

Workflow

1. Personal profile posts

1. Ask for (or pin down) the goal: reach, authority, or leads, plus the voice of the account.
2. Pick the format from references/post-formats.md based on goal and follower size: text-only for personal profiles under ~5k followers, carousels for reach, multi-image for engagement, video for demos. Do not default to the same format every time.
3. Draft the post: Hook -> Body -> CTA.

• Hook: first line under 40 characters, a complete thought, no hedging, no preamble. Numbers work. "How I [did X]" beats "How to [do X]".
• Body: 100-300+ words minimum (under 100 words is the weakest bucket for every format). Short paragraphs, sentences under 12 words, white space between lines. At least one specific claim with a number or named entity.
• CTA: end with a specific question, not "Thoughts?". Specific questions lift engagement 20-40%.

1. Run the linter: python3 bin/check_draft.py <draft-file>. Fix every FAIL, re-run until clean.
2. Post-day routine: 15-20 minutes of commenting before publishing; reply to every comment within 1-2 hours after.

2. Comment strategy

1. Build a target list: peers (similar size), aspirational accounts (~10x bigger), prospects, and the account's own commenters. See references/comment-strategy.md.
2. Write each comment as a mini-post: 2-4 sentences, one of these formulas: agree + add a personal example or stat; summarize the insight + ask a follow-up; share a quick lesson + invite replies. Never "Great post!" and never a generic compliment.
3. Daily target: 10-20 substantive comments on posting days. Comment within 60 minutes of a post going live for 3x the response rate.
4. Track what converts: which comments drive profile visits and follows, and double down on those targets and formulas.

3. Company page ops

1. Check roles before anything else: Super admin, Content admin, Analyst. Only Super admin can edit the page or manage admins; Super and Content admins can post, invite connections, and send employee notifications. See references/company-page-ops.md.
2. Growth levers, in order: employee advocacy (employee shares travel further than page organic reach), admin invites to personal connections, posting as the page itself in industry conversations, 3-5 posts per week of native formats.
3. Company page content mix: native documents/carousels (7.00% engagement rate, the best format), multi-image posts second, video native with captions, polls sparingly. External links stay off the feed: posts with outbound links lose ~60% reach, and the "link in comments" dodge is detected too.
4. Review page analytics weekly against one chosen objective (follower growth, leads, or website clicks), not all three at once.

Output Contract

A post draft ships as: the post text (copy-ready), the chosen format and why, the hook quoted separately so the requester sees it works alone, and the linter result. A comment draft ships as the comment text plus the target and formula used. A company page plan ships as a dated weekly calendar: which days, which format, which mix rule (4-1-1 or 5-3-2), and which analytics check closes the loop.

Operating Rules

1. Every factual claim in a draft (numbers, dates, quotes, studies) must be verifiable or cut. Unverifiable goes in the bin or gets reframed as opinion. No confident unverified claims, ever.
2. Never invent engagement data. Cite real studies (Metricool 2026, Socialinsider 2026, AuthoredUp 2025-26) or say "directional."
3. No external links in post bodies, no "link in comments!!" workaround language, no engagement bait, no comment pods. LinkedIn's algorithm penalizes all four.
4. No em dashes anywhere in LinkedIn copy. They are the most recognizable AI-text signal on the platform.
5. No announcement-frame openers ("We're excited to share", "Proud to announce", "Today I wanted to share"). Start with the claim.
6. No confident predictions about algorithm behavior beyond what sources establish. State the source when a recommendation rests on data.