Linkedin_mcp_custom_server

This LinkedIn MCP Server provides an interface to control LinkedIn via AI agents using the MCP framework. It exposes a suite of tools that enable authentication, content creation and management, profile retrieval, and company/job/people search, all through natural language commands. Core capabilities include secure OAuth 2.0 authentication, posting text and image content, updating and deleting posts, commenting and managing comments, and retrieving profile information. It also supports company research, job discovery, and people search to help agents interact with LinkedIn data programmatically. The available tools are designed for integration with Claude Desktop or other MCP clients, enabling you to issue high-level prompts and receive structured results such as URNs, post IDs, and profile data.

To use the server, install it via uvx (recommended) or as a Python package, then configure your MCP client to launch the server. The provided tool list shows which actions you can perform, such as generating OAuth URLs, exchanging authorization codes for access tokens, creating posts (text or image), updating or deleting posts, commenting, and querying for profiles, companies, jobs, and people. The integration pattern follows the MCP paradigm: you request a capability, the server performs the action against LinkedIn on your authenticated account, and returns structured results suitable for downstream agents or UI displays.

Prerequisites:

• Python 3.10+ and pip
• Optional: uvx for the fastest startup (recommended)
• Git for cloning from source

Option A: Install via uvx (recommended)

1. Ensure uvx is installed. If not, install uvx from your environment’s package manager.
2. Run the server:

uvx linkedin-mcp-server

Option B: Install from PyPI (pip)

1. Ensure Python 3.10+ is installed.
2. Install the package from PyPI:

pip install linkedin-mcp-server

3. Run the server (if not using uvx):

python -m linkedin_mcp_server

Option C: Install from source (development or customization)

1. Clone the repository:

git clone https://github.com/SARAMALI15792/LinkedIn_mcp_custom_server.git

2. Install in editable mode:

cd LinkedIn_mcp_custom_server
pip install -e .

3. Run the server (example):

python -m linkedin_mcp_server

Environment preparation (required for OAuth):

• Create a .env file with:

LINKEDIN_CLIENT_ID=your_client_id_here
LINKEDIN_CLIENT_SECRET=your_client_secret_here
LINKEDIN_REDIRECT_URI=http://localhost:8000

• Ensure these values are loaded by your runtime environment or explicitly exported when launching the MCP server.

Tips and common issues:

• Ensure the LinkedIn app is configured with the correct redirect URI (http://localhost:8000) used during the OAuth flow.
• Never commit your .env file containing client secrets; add it to your .gitignore.
• If using Claude Desktop, you may configure the server differently depending on whether you installed the package via uvx or pip; adjust the command in Claude's mcpServers accordingly.
• When using the Python runtime, you may need to set environment variables in your editor or runtime to ensure the OAuth credentials are accessible to the server.
• For production usage, consider securing the server behind a gateway and rotating access tokens periodically according to LinkedIn’s policies.