chuk-mcp-linkedin

A comprehensive design system MCP server for creating LinkedIn posts with shadcn-inspired component architecture, CVA-style variants, and powerful theming.

Overview

chuk-mcp-linkedin brings design system principles to LinkedIn content creation. Create posts using composable components, variants, and themes - similar to modern frontend design systems like shadcn/ui but for social media content.

Features

Component-Based Architecture: 13+ specialized post types (text, document, poll, video, carousel, etc.)
Variant System: CVA-inspired variants with compound variant support
Theme System: 10 pre-built themes (thought leader, storyteller, community builder, etc.)
Composition Patterns: Build complex posts from subcomponents (hooks, body, CTA, hashtags)
Design Tokens: Research-backed tokens for engagement, formatting, and timing
2025 Performance Data: Built-in optimization based on 1M+ post analysis
MCP Integration: Full Model Context Protocol support for LLM workflows

2025 LinkedIn Performance Insights

Based on analysis of 1M+ posts across 9K company pages:

Top Performing Formats

Document Posts (PDF) - 45.85% engagement rate (HIGHEST)
Poll Posts - 200%+ higher reach (MOST UNDERUSED)
Video Posts - 1.4x engagement (rising fast)
Image Posts - 2x more comments than text
Carousel Posts - Declining (keep to 5-10 slides)

Key Insights

Polls achieve highest reach but are least used (opportunity!)
Document posts dominate engagement (carousel's successor)
Video usage up 69%, vertical format preferred
First 210 characters critical (before "see more")
First hour engagement determines algorithmic reach

Installation

pip install chuk-mcp-linkedin

Quick Start

Simple Text Post

from chuk_mcp_linkedin import LinkedInManager, ThemeManager # Initialize manager = LinkedInManager() theme_mgr = ThemeManager() # Create thought leadership post theme = theme_mgr.get_theme("thought_leader") post = manager.create_text_post( commentary="""80% of B2B decision makers prefer thought leadership content over ads. Yet most companies just promote. Here's what actually works: → Lead with insights, not products → Share frameworks, not features → Tell stories, not sales pitches → Build trust, not transactions The algorithm rewards value.""", variant="insight", tone="professional", theme=theme ) post.publish(visibility="PUBLIC")

Document Post (Highest Engagement)

from chuk_mcp_linkedin import DocumentPost, ChartComponents # Create document post doc = DocumentPost( commentary="Our Q4 results are in. Here's what we learned 📊", variant="report", theme=theme_mgr.get_theme("data_driven") ) # Add slides doc.add_slide( layout="title_slide", content={"title": "Q4 2024 Results", "subtitle": "Growth & Insights"} ) # Add metrics with chart metrics_chart = ChartComponents.metric_grid( title="Key Metrics", metrics=[ {"label": "Revenue", "value": "$1.2M", "trend": "+12%"}, {"label": "Customers", "value": "450", "trend": "+25%"}, ] ) doc.add_slide( layout="content_slide", content={"title": "Q4 Performance", "chart": metrics_chart} ) doc.publish()

Poll Post (Highest Reach)

# Create poll (200%+ higher reach!) poll = manager.create_poll_post( commentary="Quick question for my network:\n\nWhat's your biggest LinkedIn challenge in 2025?", question="Pick your top challenge:", options=[ "Creating consistent content", "Growing engagement", "Converting leads", "Building community" ], duration_days=3, purpose="research", theme=theme_mgr.get_theme("community_builder") ) poll.publish()

Composition Pattern (Advanced)

from chuk_mcp_linkedin import ComposablePost # Build post using composition post = (ComposablePost("text", theme=theme) .add_hook("stat", "95% of LinkedIn posts get zero comments") .add_body(""" Here's why (and how to fix it): Most posts lack these 3 elements: → Strong hook (first 210 characters) → Clear value (what's in it for them) → Conversation starter (invite engagement) Start treating posts like conversations, not broadcasts. """, structure="listicle") .add_cta("curiosity", "What's your biggest LinkedIn frustration?") .add_hashtags(["LinkedInTips", "ContentStrategy"]) .optimize_for_engagement() .compose()) manager.publish_text(post)

Component System

Post Types

TextPost - Simple text updates with variants (story, insight, question, listicle, hot_take)
DocumentPost - PDF carousels (highest engagement format in 2025)
PollPost - Polls for engagement (highest reach - 200%+ above average)
VideoPost - Video content (1.4x engagement)
CarouselPost - Multi-image storytelling (keep under 10 slides)
ImagePost - Single image posts
StoryPost - Personal narratives with emotional arc
ArticlePost - Link sharing with commentary
Plus: BehindTheScenesPost, HumorPost, AnnouncementPost, EventPost, ThoughtLeadershipPost

Subcomponents

Hook - Opening hooks (question, stat, story, controversy, list, curiosity)
Body - Main content with structures (linear, listicle, framework, story_arc, comparison)
CallToAction - Engagement drivers (direct, curiosity, action, share, soft)
Hashtags - Hashtag strategies (minimal, optimal, branded, trending)

Themes

10 pre-built themes for different LinkedIn personas:

thought_leader - Establish expertise and authority
personal_brand - Build authentic personal connection
technical_expert - Deep technical knowledge
community_builder - Foster conversation and connection
corporate_professional - Polished corporate communication
contrarian_voice - Challenge status quo, spark debate
storyteller - Narrative-driven emotional connection
data_driven - Let numbers tell the story
coach_mentor - Guide and support your audience
entertainer - Make LinkedIn fun and memorable

MCP Server Tools

Draft Management

linkedin_create - Create new draft
linkedin_list - List all drafts
linkedin_switch - Switch between drafts
linkedin_delete - Delete draft
linkedin_get_info - Get draft details

Content Creation

linkedin_text_post - Create text post
linkedin_document_post - Create document post
linkedin_poll_post - Create poll post
linkedin_video_post - Create video post
linkedin_carousel_post - Create carousel post

Composition

linkedin_add_hook - Add opening hook
linkedin_add_body - Add main content
linkedin_add_cta - Add call-to-action
linkedin_add_hashtags - Add hashtags

Enhancement

linkedin_optimize_length - Optimize text length
linkedin_suggest_emojis - Suggest emoji placement
linkedin_format_for_scannability - Add formatting for readability
linkedin_apply_theme - Apply theme to draft

Publishing

linkedin_publish - Publish to LinkedIn
linkedin_schedule - Schedule for later
linkedin_preview - Get preview
linkedin_export_draft - Export as JSON

Analytics

linkedin_get_post_stats - Get post analytics
linkedin_get_suggestions - Get content suggestions
linkedin_analyze_draft - Analyze draft performance potential

Design Tokens

Research-backed design tokens for optimal engagement:

Text Formatting

Character limits (3000 max, 210 before "see more")
Ideal lengths (micro: 50-150, short: 150-300, medium: 300-800, long: 800-1500, story: 1000-3000)
Line break styles (dense, readable, scannable, dramatic, extreme)

Emoji Usage

None, minimal (1%), moderate (5%), expressive (10%), heavy (15%)

Hashtags

Optimal count: 3-5 hashtags
Placement strategies: inline, mid, end, first_comment

Engagement Patterns

Hook types with power ratings (controversy: 0.95, stat: 0.9, story: 0.85)
CTA styles (direct, curiosity, action, share, soft)
First hour targets (minimum: 10, good: 50, viral: 100 engagements)

Timing

Best days: Tuesday, Wednesday, Thursday
Best hours: 7-9 AM, 12-2 PM, 5-6 PM
Optimal frequency: 4-5 posts per week

Architecture

src/chuk_mcp_linkedin/ ├── components/ # Post type components ├── subcomponents/ # Composition subcomponents ├── tokens/ # Design tokens ├── themes/ # Theme system ├── layouts/ # Visual layouts ├── charts/ # Chart components ├── variants.py # Variant system ├── composition.py # Composition patterns ├── registry.py # Component registry ├── manager.py # Draft management └── server.py # MCP server

Development

# Install dev dependencies pip install -e ".[dev]" # Run tests pytest # Format code black src tests # Type checking mypy src

License

MIT

Credits

Built by Christopher Hay

Based on 2025 LinkedIn performance data from analysis of 1M+ posts across 9K company pages.

Design system principles inspired by shadcn/ui and CVA.

LinkedIn Content Creation MCP Server