A Claude Code plugin that drafts LinkedIn posts, profiles, carousels, DMs, and comments in your actual voice. Refuses templates, refuses scaled outreach, refuses AI vocabulary — by design.
Each skill is a self-contained SKILL.md file. Claude Code auto-loads them by trigger phrase. No commands to remember.
Set up once. Every other skill reads what these produce.
train-voice01Reads samples of your past writing — LinkedIn data export, blog, pasted posts — and produces voice-profile.md that every downstream writing skill reads automatically.
audit-profile02Diagnoses your current LinkedIn profile section by section. Severity-rated issues, prioritized fix list. Run before rewrite-profile.
rewrite-profile03Rewrites the full profile as one positioning unit: headline, About, Featured, Experience bullets, Skills. Not a section-by-section pass.
plan-content04Defines 3-5 content pillars, audience, and cadence. Outputs content-plan.md that analyze-performance and review-post read later.
The daily writing work. Posts, hooks, carousels, scripts.
find-ideas01Interviews you to surface specific, postable ideas. Outputs 2-5 idea seeds with concrete anchors and which next skill to use.
write-hook02Generates 3-5 hook variants that fit the 210-character LinkedIn see-more fold, with the strongest variant marked and reasoning.
write-post03Drafts a 1,000-1,300 character text post in your voice. The workhorse skill. Anti-AI-tell rules embedded inline, not as a post-hoc pass.
build-carousel04Plans and writes a 6-12 slide document carousel: copy plus visual brief. Outputs a spec you assemble into a PDF in Canva or Figma.
write-longform05Writes a 600-2,000 word article or newsletter issue meant to live permanently on your profile rather than in the feed.
write-video-script06Writes a 30-90 second video script with beats, timing, and on-screen text. For ideas with visual or demonstrative power text cannot capture.
repurpose-content07Takes one piece of source content — talk, podcast, blog, doc — and produces 3-7 distinct LinkedIn pieces from different angles, not 7 versions of the same angle.
One-to-one and thread work. No mail-merge. No engagement bait.
write-comment01High-signal comments on others' posts. Comments now carry ~15x the algorithmic weight of likes. No great-post-at-scale templates.
write-dm02Connection requests, cold outreach, follow-ups, replies. One message at a time. No mail-merge. No 100-DM blast.
write-recommendation03LinkedIn recommendations for former colleagues, with specific anchors instead of templated praise. Avoids the standard recommendation cadence.
Close the feedback cycle. See what landed and adjust the plan.
analyze-performance01Clusters posts by pillar, format, hook type, and length using the free LinkedIn data export CSV. Identifies what is working and what is not.
review-post02Postmortem on one specific post: hook, body, close, format, timing. Three lessons for the next attempt, not five.
The humanizer pattern library every writing skill reads from.
humanizer01Canonical 33-pattern reference built on Wikipedia's Signs of AI writing (WikiProject AI Cleanup). The 10 writing skills embed the same rules inline; this one is invoked when humanizing pasted external text.
They generate hooks from templates, evade AI detection, and produce posts in the same voice as 50,000 other posts the same week. The result is the feed everyone scrolls past.
LinkedIn's 2026 ranking rewards dwell time, substantive comments, and content that earns the “see more” click. The way to earn those is not better templates — it's starting from something specific in your voice and producing clean, human-sounding text on the first pass.
No hook-value-CTA. No PAS, AIDA, no 5-lessons-I-learned. LinkedIn's 2026 ranking now downranks these patterns.
The DM skill writes one good message at a time, not 100 mail-merged ones.
Comments add substance, not great-post-at-scale. No follow-for-follow.
Anti-AI-tell rules are baked into drafting itself. The first output is the clean output.
$ /plugin marketplace add warpirate/linkedin-maxxing$ /plugin install linkedin-maxxing@linkedin-maxxing-marketplaceRestart the Claude Code session after install. Skills auto-load based on description trigger phrases — no commands to invoke manually.
Each skill is a self-contained SKILL.md under plugins/linkedin-maxxing/skills/<name>/. Drop the directory wherever your client reads skills.
Every plugin is MIT licensed and lives in a public GitHub repo. Stars help the catalog reach the next person who needs it.
Last commit 12h ago on main
Sending 100 personalized DMs? Not this tool.
No comment generators. No follow-for-follow.
If you have nothing to say, the skills ask for material first.
Produces work worth posting. What happens after isn't guaranteed.
Read-side skills use the free LinkedIn data export.
You stay in the loop. Skills draft and critique. You publish.