Profile-driven LinkedIn content generation plugin for Claude Code, Claude Desktop, and Cowork. Built on a comprehensive 2025-2026 LinkedIn knowledge base with built-in defenses against AI slop. https://github.com/juliandickie/linkedin-post-studio
Find a file
Julian Dickie 7f64429a9a fix: rename self-marketplace idd-plugins -> linkedin-post-studio-marketplace
The self-marketplace declared name 'idd-plugins' - the collision-prone
legacy name retired elsewhere in the ecosystem (scribe-marketplace,
creators-studio-marketplace, spiffy). Renamed for consistency and to
fully kill the idd-plugins landmine. Source './' was already valid and
is unchanged.
2026-05-16 16:18:40 +10:00
.claude-plugin fix: rename self-marketplace idd-plugins -> linkedin-post-studio-marketplace 2026-05-16 16:18:40 +10:00
assets feat: add marketplace catalog and plugin icon for one-line install 2026-04-27 08:24:48 +10:00
skills feat: migrate 15 slash commands from legacy commands/ format to skills/<name>/SKILL.md 2026-04-20 02:22:41 +10:00
CHANGELOG.md feat: add marketplace catalog and plugin icon for one-line install 2026-04-27 08:24:48 +10:00
icon.png feat: add marketplace catalog and plugin icon for one-line install 2026-04-27 08:24:48 +10:00
LICENSE chore: add MIT LICENSE to plugin 2026-04-19 23:02:25 +10:00
README.md feat: add marketplace catalog and plugin icon for one-line install 2026-04-27 08:24:48 +10:00

LinkedIn Post Studio icon

LinkedIn Post Studio

Profile-driven LinkedIn content generation built on a comprehensive 20252026 LinkedIn knowledge base. Generates text posts, carousels, hooks, polls, video scripts, newsletters, calendars, companion comments, and more — in YOUR voice, with built-in defenses against AI slop.

What it does

  • 12 output types — text posts, carousels, hook variations, companion comments, polls, video scripts, newsletters, content calendars, QA reviews, idea mining, content repurposing, post teardowns
  • Profile-driven — captures your voice once (6 methods including macOS/iPhone dictation, Whisper, paste), reuses across every generation
  • Industry-aware — 12 shipped industry modules with voice defaults and compliance rules (healthcare HIPAA, legal ABA, finance SEC/FINRA, and more)
  • Anti-AI-slop — every draft passes through banned-vocabulary, banned-structure, and required-element checks before output
  • Local-first — your data stays on your machine; no analytics, no cloud sync (unless you point data_location at a synced folder)

Why it's different

  • Voice preservation over templates — we capture how you actually write; we don't apply a template to your topic
  • Compliance built in — regulated industries get proper disclosures; no retrofitting disclaimer text
  • Always-on QA — every generation runs a pre-publish checklist regardless of workflow; nothing ships without passing

Where it works

Surface Skill (natural language) Full plugin (slash commands)
Claude Code (terminal)
Claude Desktop — code mode
Claude Desktop — chat mode
Claude Web (claude.ai)
Cowork

Full plugin gives you 15 slash commands and profile management UX. Skill-only gives you full generation capability via natural language — just say "write me a LinkedIn post about X."

Installation

This repo ships its own Claude Code plugin marketplace, so you get one-line install and one-line updates:

# Inside Claude Code (or Desktop code mode / Cowork)
/plugin marketplace add juliandickie/linkedin-post-studio
/plugin install linkedin-post-studio@idd-plugins

To update later:

/plugin marketplace update idd-plugins
/plugin update linkedin-post-studio

Both Claude Code commands also work in non-interactive mode from your shell:

claude plugin marketplace add juliandickie/linkedin-post-studio
claude plugin install linkedin-post-studio@idd-plugins

Option B — Manual git clone (Claude Code / Desktop code mode / Cowork)

If you'd rather manage the install with git directly:

cd ~/.claude/plugins/
git clone https://github.com/juliandickie/linkedin-post-studio.git

All 15 slash commands and the skill are auto-discovered. To update, cd into the cloned directory and git pull.

Option C — Skill only (Claude Web / Desktop chat mode)

# Copy the skill directory
cp -r linkedin-post-studio/skills/linkedin-post-studio ~/.claude/skills/

# Or symlink for automatic updates when you git pull
ln -sfn "$(pwd)/linkedin-post-studio/skills/linkedin-post-studio" ~/.claude/skills/linkedin-post-studio

No slash commands, but the skill triggers on natural language in any Claude surface.

Option D — Project-scoped (Claude Code only)

For users who want the plugin only for one project (e.g., an agency working on a single client):

cd /path/to/your-project
mkdir -p .claude/plugins/
git clone https://github.com/juliandickie/linkedin-post-studio.git .claude/plugins/linkedin-post-studio

Or, project-scoped via the marketplace (auto-prompts teammates to install it when they trust the project):

// .claude/settings.json
{
  "extraKnownMarketplaces": {
    "idd-plugins": {
      "source": { "source": "github", "repo": "juliandickie/linkedin-post-studio" }
    }
  },
  "enabledPlugins": {
    "linkedin-post-studio@idd-plugins": true
  }
}

⚠️ Restart required after install

Claude loads plugins and skills at startup, not on the fly. After installing (or updating) the plugin, restart your Claude surface so the 15 slash commands and the skill show up:

  • Claude Desktop — fully quit the app (⌘Q on macOS) and relaunch
  • Claude Code (terminal) — exit your current session and start a new one
  • Claude Web — reload the browser tab
  • Cowork — restart the Cowork session

If you install the plugin and type /linkedin- and see no auto-complete suggestions, you haven't restarted yet. After restart, typing /linkedin- should show all 15 commands, and natural language like "write me a LinkedIn post about pricing strategy" should trigger the orchestrator.

First-time setup (5 minutes)

On first invocation, the skill walks you through onboarding:

  1. Identity — name, profile type (personal / company / client), industry
  2. Audience — primary role, pain points, sophistication level
  3. Goals — default goal (lead-gen / thought-leadership / brand-awareness / community)
  4. Compliance — only runs for regulated industries; loads defaults and lets you customize
  5. Voice capture (optional, deferrable) — 15 min to capture your voice from past posts; defer with a quality warning if you prefer

The onboarding offers 6 voice-capture methods (see below).

Commands

Command Purpose
/linkedin-post-studio Show status, profiles, recent drafts, full command menu
/linkedin-onboard Create a new profile or re-onboard an existing one
/linkedin-profile Manage profiles (list, switch, edit, delete, view)
/linkedin-post Generate a text post (8001,000 chars)
/linkedin-carousel Generate a carousel brief (610 slides)
/linkedin-hooks Generate 15 hook variations mixing categories
/linkedin-companion Generate 5 companion comment prompts on a theme
/linkedin-poll Generate a poll (3 options + "Other", 7-day)
/linkedin-video Generate a 6090s video script with caption plan
/linkedin-newsletter Generate a newsletter edition + promo post
/linkedin-calendar Generate a content calendar (week or month)
/linkedin-qa Run QA checklist on an existing draft
/linkedin-mine Mine post angles from a source (Reddit, reviews, transcripts)
/linkedin-repurpose Convert long-form content into multiple LinkedIn assets
/linkedin-teardown Analyze a viral post + generate your version

Voice capture methods

The more samples, the better the output. Aim for 5+ samples; 1015 is ideal.

  1. Paste posts directly in chat — simplest; paste 315 past posts separated by ---
  2. Folder path — point at a folder of .md or .txt files; the skill reads all of them
  3. iPhone / iPad dictation via Notes — 5-minute voice memo in Notes, paste the transcript
  4. Descript / Otter / Fireflies — record, export transcript, paste
  5. Voice memo + Whisper — record on phone, run through MacWhisper or whisper.cpp, paste
  6. macOS native dictation — System Settings → Keyboard → Dictation → enable; speak directly into any text field

Supported industries

Out-of-the-box compliance and voice defaults for:

  • B2B SaaS — founder-led narratives, specific metrics, case studies
  • Professional Services — consulting, accounting, agencies
  • Creator Economy — transformation stories, revenue screenshots, cohort launches
  • E-commerce / DTC — operator-to-operator content, CAC/LTV math, creative teardowns
  • Healthcare (regulated) — HIPAA, FDA
  • Legal (regulated) — ABA Model Rules, state variants
  • Finance (regulated) — SEC Marketing Rule, FINRA 2210
  • Real Estate (regulated) — Fair Housing, RESPA
  • Manufacturing (regulated) — OSHA, ITAR
  • Energy & Utilities (regulated) — FERC, NERC CIP
  • Dental Education (IDD-style, regulated-adjacent)
  • Generic (fallback for any other industry)

Adding a new industry: copy references/industry/industry-template.md and fill in. See docs/adding-industries.md.

Where your data lives

Data Location Purpose
App config ~/.claude/data/linkedin-post-studio/config.md Default profile pointer, feature flags, data_location override
Profiles ~/Documents/LinkedIn Post Studio/profiles/ Per-profile markdown files
Voice samples ~/Documents/LinkedIn Post Studio/voice-samples/<profile>/ Past posts the skill reads for voice matching
Drafts ~/Documents/LinkedIn Post Studio/drafts/ Every generated draft
Archive ~/Documents/LinkedIn Post Studio/archive/ Posted drafts (move manually or via command)

Your data is never sent anywhere outside your Claude session. No analytics, no tracking, no cloud sync (unless you point data_location at a synced folder for team sharing).

Roadmap

See docs/roadmap.md for the full roadmap with status.

Highlights:

  • v2 — Creators-studio integration (images and videos from posts), pre-publish hook, scheduling
  • v3 — Automated performance tracking, Apify-based competitive research, performance dashboard
  • v4 — Multi-language support, team/agency profiles, custom pillar frameworks

Explicitly NOT on the roadmap: engagement pods, third-party DM automation tools (ToS violations), auto-publishing without user review, "guaranteed viral" promises.

Contributing

Design spec: specs/2026-04-19-linkedin-post-studio-design.md.

License

MIT — see LICENSE.