Meridian LinkedIn MCP

Elegant Software Solutions

Production-grade Model Context Protocol (MCP) server bridging Claude Code with LinkedIn's professional networking platform. Enables AI assistants to access profiles, create posts and polls, manage social engagement, and handle organizations programmatically. Features OAuth 2.0 3-legged flow with automatic token refresh and secure file-based credential storage.

3K+

Lines of Code

MCP Tools

MCP Resources

Python Files

MCP & OAuth Architecture

Contribution as Principal Architect

Designed provider abstraction isolating LinkedIn API complexity from MCP layer
Created token management strategy with caching and 5-minute refresh buffer
Established security-first credential storage (0600 file permissions, no env vars)
Built data normalization pipeline for consistent API response schemas

Core Capabilities

Profile Access

Multi-source profile aggregation from /me, /userinfo, and /emailAddress endpoints into unified view.

Content Creation

Post creation with automatic hashtag formatting (#AI → {hashtag|#|AI}) and poll support with 2-4 options.

Social Engagement

Reactions (LIKE, CELEBRATE, SUPPORT, LOVE, INSIGHTFUL, FUNNY) and comments on posts.

Secure Authentication

File-based credential storage with 0600 permissions, no environment variable exposure.

Auto Token Refresh

5-minute buffer pre-refresh preventing edge-case expiry with graceful fallback.

Organization Management

Post on behalf of administered company pages with organization context.

Modules & Services

MCP Server

FastMCP-based server exposing 18 tools for profile access, post management, social interactions, and authentication. Declarative tool definitions with comprehensive parameter validation.

OAuth 2.0 Handler

Complete 3-legged OAuth flow with CSRF protection via state parameter, browser-based authorization, HTTP callback server, and beautiful success/error UI.

Token Manager

Intelligent token lifecycle management with in-memory caching, 5-minute buffer pre-refresh, graceful fallback on refresh failure, and secure file-based storage (0600 permissions).

LinkedIn Provider

High-level API wrapper (1,211 lines) with data normalization pipeline transforming nested LinkedIn responses into flat, consistent schemas. Handles hashtag auto-formatting and multi-source profile aggregation.

Tech Stack

Languages

Python 3.11+

Frameworks

FastMCPPydantic 2pydantic-settings

Infrastructure

LinkedIn API v202501

Data

Loading project...