LinkedIn Content Creation MCP Server

# LinkedIn MCP Server Examples Clean, focused examples demonstrating core functionality of the LinkedIn MCP Server. ## Important Note **All examples use OAuth 2.1 for authentication.** The legacy examples that used `LINKEDIN_ACCESS_TOKEN` have been removed. For OAuth setup, see: - **[oauth_linkedin_example.py](#5-oauth--authentication)** - Complete OAuth flow demonstration - **[docs/OAUTH.md](../docs/OAUTH.md)** - OAuth setup guide ## Available Examples ### 1. **Complete Component Showcase** ⭐ #### `showcase_all_components.py` **THE MAIN SHOWCASE** - Interactive HTML previews of EVERY post component. **Features:** - **Post Structure:** Hook, Body, CTA, Hashtags, Separator - **Charts:** Bar, Metrics, Comparison, Progress, Ranking (5 types) - **Features:** Quote, BigStat, Timeline, KeyTakeaway, ProCon (5 types) - **Combined Examples:** Multiple components in real-world posts - Generates 13 HTML previews - LinkedIn-style formatting - Opens in browser **Run:** ```bash python examples/showcase_all_components.py ``` **Output:** Saves 13 HTML files to `~/.linkedin_drafts/previews/showcase/` This is the best place to start to see everything the system can do! --- ### 2. **Charts & Data Visualization** #### `demo_atomic_charts.py` Terminal output demonstration of all chart components with Pydantic validation. **Features:** - Bar charts with colored emoji squares - Metrics charts with ✅/❌ indicators - Comparison charts (A vs B) - Progress bars (0-100%) - Ranking charts with medals - Combined charts in one post - Pydantic data validation examples **Run:** ```bash python examples/demo_atomic_charts.py ``` #### `demo_charts_preview.py` Interactive HTML previews of all chart and feature components. **Features:** - All chart types with LinkedIn-style previews - Feature components (Quote, BigStat, Timeline, KeyTakeaway, ProCon) - Combined components example - Opens previews in browser - Interactive preview selection **Run:** ```bash python examples/demo_charts_preview.py ``` ### 3. **Media Attachments** #### `showcase_media_types.py` Demonstrates different media types in LinkedIn posts. **Features:** - Single image posts - Multiple images (2-4 grid layout) - Video posts with thumbnails - PDF document attachments - PowerPoint presentations - Interactive HTML previews **Run:** ```bash python examples/showcase_media_types.py ``` ### 4. **Documents** #### `demo_document_page_preview.py` Shows document-to-image conversion for preview (like LinkedIn does). **Features:** - PDF pages rendered as images - PowerPoint slides rendered as images - Interactive carousel navigation - Cached conversions for performance - Requires: `pip install chuk-mcp-linkedin[preview]` **Run:** ```bash python examples/demo_document_page_preview.py ``` **Requirements:** - Install preview dependencies: `pip install chuk-mcp-linkedin[preview]` - Install poppler: `brew install poppler` (macOS) or `sudo apt-get install poppler-utils` (Ubuntu) #### `demo_document_upload_and_attach.py` Complete workflow for LinkedIn's Documents API. **Features:** - Document validation (size, format, pages) - Upload documents to LinkedIn API - Get document URNs - Create posts with document attachments - Integration with ComposablePost **Run:** ```bash python examples/demo_document_upload_and_attach.py ``` **Note:** Requires LinkedIn API credentials to actually upload. Example shows the structure. ## Quick Start ### Install Dependencies **Core functionality:** ```bash pip install chuk-mcp-linkedin ``` **With preview support:** ```bash pip install "chuk-mcp-linkedin[preview]" brew install poppler # macOS only, for PDF support ``` ### Run Any Example ```bash # From the project root python examples/demo_charts_preview.py # Or from examples directory cd examples python demo_charts_preview.py ``` ## Example Categories ### Complete Showcase ⭐ - `showcase_all_components.py` - **START HERE** - Every component with HTML preview ### Charts & Visualization - `demo_atomic_charts.py` - Terminal output with validation - `demo_charts_preview.py` - HTML previews in browser ### Media Types - `showcase_media_types.py` - Images, videos, documents ### Documents - `demo_document_page_preview.py` - Document rendering - `demo_document_upload_and_attach.py` - LinkedIn API workflow ## Output Locations Examples save previews to: - **HTML previews:** `~/.linkedin_drafts/previews/` - **Document cache:** `~/.linkedin_drafts/document_cache/` ## Common Patterns ### Create a Post with Charts ```python from chuk_mcp_linkedin.posts import ComposablePost post = ComposablePost("text") post.add_hook("stat", "📊 Data-Driven Insights") post.add_bar_chart( data={"Q1": 100, "Q2": 150, "Q3": 200}, title="QUARTERLY GROWTH", unit="customers" ) post.add_cta("curiosity", "What's your growth rate?") post.add_hashtags(["Data", "Growth"]) text = post.compose() ``` ### Generate HTML Preview ```python from chuk_mcp_linkedin.preview import LinkedInPreview draft_data = { "name": "My Post", "post_type": "text", "content": {"composed_text": text} } html = LinkedInPreview.generate_html(draft_data) preview_path = LinkedInPreview.save_preview(html, "my_post.html") ``` ### Publishing to LinkedIn Publishing requires OAuth authentication. See the OAuth example: ```python # See examples/oauth_linkedin_example.py for complete OAuth flow # The MCP server handles OAuth automatically when configured properly ``` For production use, configure the MCP server with OAuth: ```bash export LINKEDIN_CLIENT_ID=your_client_id export LINKEDIN_CLIENT_SECRET=your_client_secret export SESSION_PROVIDER=memory uv run linkedin-mcp http --port 8000 ``` ## Testing Examples All examples are self-contained and can be run independently. They use test files from: ``` /test_files/ ├── test_image_1.png ├── test_document.pdf └── test_presentation.pptx ``` ## Need Help? - **Documentation:** `/docs/ARCHITECTURE.md` - **API Reference:** `/docs/README.md` - **Issues:** https://github.com/chrishayuk/chuk-mcp-linkedin/issues ## Example Output All examples provide clear terminal output and/or open HTML previews in your browser: - ✅ Chart examples show formatted text output - 🌐 Preview examples open HTML in browser - 📊 Document examples show conversion progress - 🔗 API examples show workflow structure Happy posting! 🚀 --- ## 5. **OAuth & Authentication** ### `oauth_linkedin_example.py` Complete LinkedIn OAuth flow demonstration with chuk-sessions token storage. **Features:** - MCP client registration - LinkedIn OAuth authorization flow - Token exchange and validation - Token refresh with rotation - chuk-sessions storage (memory or Redis) - Step-by-step flow visualization **Run:** ```bash # Set LinkedIn credentials export LINKEDIN_CLIENT_ID=your_client_id export LINKEDIN_CLIENT_SECRET=your_client_secret # Run with memory backend (default) python examples/oauth_linkedin_example.py # Or with Redis backend export SESSION_PROVIDER=redis export SESSION_REDIS_URL=redis://localhost:6379/0 python examples/oauth_linkedin_example.py ``` **Output:** ``` ================================================================================ LinkedIn OAuth Flow Example with chuk-sessions ================================================================================ 📦 Token Storage Backend: MEMORY Step 1: Registering MCP Client -------------------------------------------------------------------------------- ✅ Client registered successfully! Client ID: xoUmSxhdFGnoRVQ-mz4i8g Client Secret: 5c7UhwF0KG... Step 2: Initiating LinkedIn Authorization -------------------------------------------------------------------------------- 🔗 LinkedIn authorization required ... 📊 Complete OAuth Flow: ──────────────────────────────────────────────────────────────────────────── 1. MCP Client → OAuth Server: Request authorization 2. OAuth Server → LinkedIn: Redirect user for login 3. User authorizes on LinkedIn 4. LinkedIn → OAuth Server: Callback with code 5. OAuth Server → LinkedIn: Exchange code for token 6. OAuth Server: Store LinkedIn token in chuk-sessions 7. OAuth Server → MCP Client: Return authorization code 8. MCP Client → OAuth Server: Exchange code for MCP token 9. MCP Client uses MCP token for API calls 10. OAuth Server validates MCP token & uses LinkedIn token ──────────────────────────────────────────────────────────────────────────── ``` **Prerequisites:** 1. Create LinkedIn app at https://www.linkedin.com/developers/apps 2. Add redirect URI: `http://localhost:8000/oauth/linkedin/callback` 3. Copy Client ID and Client Secret **Token Storage:** Tokens are automatically stored in chuk-sessions with TTL: - Authorization codes: 10 minutes - MCP access tokens: 1 hour - MCP refresh tokens: 30 days - LinkedIn tokens: 60 days **Backend Options:** | Backend | Use Case | Configuration | |---------|----------|---------------| | Memory (default) | Development, testing | No config needed | | Redis | Production, multi-user | `SESSION_PROVIDER=redis` | **Inspect Tokens (Redis):** ```bash # Connect to Redis redis-cli # List all tokens KEYS chuk-mcp-linkedin:* # Get specific token GET chuk-mcp-linkedin:access_token:abc123... # Check time remaining TTL chuk-mcp-linkedin:access_token:abc123... ``` **Full Production Setup:** ```bash # Terminal 1: Start OAuth server export LINKEDIN_CLIENT_ID=your_client_id export LINKEDIN_CLIENT_SECRET=your_client_secret export SESSION_PROVIDER=redis export SESSION_REDIS_URL=redis://localhost:6379/0 uv run linkedin-mcp http --port 8000 # Terminal 2: Configure MCP client (Claude Desktop) # Edit ~/Library/Application Support/Claude/claude_desktop_config.json: { "mcpServers": { "linkedin": { "command": "linkedin-mcp", "args": ["http", "--port", "8000"], "authorization": { "type": "oauth2", "authorization_url": "http://localhost:8000/oauth/authorize", "token_url": "http://localhost:8000/oauth/token" } } } } ``` **See Also:** - `OAUTH_WITH_CHUK_SESSIONS.md` - Complete OAuth + chuk-sessions guide - `OAUTH_IMPLEMENTATION_SUMMARY.md` - OAuth architecture details