Mcp Linkedin

What is Mcp Linkedin?

Mcp Linkedin is an MCP server that wraps the Unipile API to let AI assistants publish, comment, and react on LinkedIn. It is built for AI assistants like Claude Code and Claude Desktop, or any MCP-compatible client.

How to use Mcp Linkedin?

Install Node.js 18+, clone the repository, run npm install, then configure your MCP client (e.g., Claude Code or Claude Desktop) with the server command and environment variables UNIPILE_API_KEY and UNIPILE_DSN. The server exposes three tools: linkedin_publish, linkedin_comment, and linkedin_react. All publishing actions default to preview mode (dry_run: true) and require explicit confirmation to go live.

Key features of Mcp Linkedin

• 3 tools: publish, comment, react
• Dry run by default – preview before publishing
• Auto-likes posts immediately after publishing
• Supports media attachments (local files or URLs – images and video)
• Company @mentions auto-resolved via Unipile

Use cases of Mcp Linkedin

• An AI assistant drafts and previews LinkedIn posts before publishing
• Automated commenting on existing LinkedIn posts
• Reacting to posts with various reaction types (like, celebrate, love, etc.)
• Scheduled or batched content publishing via MCP clients

FAQ from Mcp Linkedin

What is the default behavior for publishing?

All publishing actions default to preview mode (dry_run: true). Nothing goes live without explicitly setting dry_run: false after reviewing the preview.

What are the prerequisites?

You need Node.js 18+ and a Unipile account. Unipile handles LinkedIn OAuth – you do not need to manage LinkedIn tokens directly.

How do I configure the server?

Set UNIPILE_API_KEY and UNIPILE_DSN in the MCP client configuration file (e.g., ~/.claude/mcp.json for Claude Code). No .env file is used.

Can I attach media to posts?

Yes. You can pass local file paths or URLs for jpg, png, gif, webp, and mp4 files. Files are validated and temporarily downloaded.

What transports does the server use?

Mcp Linkedin uses MCP stdio transport (JSON-RPC over standard input/output). No other transports (HTTP, WebSocket) are mentioned.