MCP quickstart

From LinkedIn saves to Claude context in four steps

ContexIn runs locally: the Chrome extension exports your saved posts as JSON, and the MCP server reads that local file so Claude, ChatGPT, or any MCP client can search it.

01 downloads

02 export

Export your saved posts

Install the Chrome extension, open LinkedIn saved posts, collect visible saves, then download the JSON export.

https://www.linkedin.com/my-items/saved-posts/

03 store

Put the JSON somewhere stable

Move the downloaded file out of Downloads so your MCP client can read the same path every time.

mkdir -p ~/.contexin
mv ~/Downloads/contexin-saved-posts.json ~/.contexin/saved-posts.json

04 run

Run the local MCP server

Download the MCP server package, install dependencies, build it, and point POSTMIND_DATA at your export.

unzip postmind-mcp-server.zip
cd postmind-mcp-server
npm install
npm run build
POSTMIND_DATA=~/.contexin/saved-posts.json node dist/index.js

05 connect

Connect Claude Desktop

Add this server entry to your Claude Desktop MCP config, replacing both paths with absolute paths on your computer.

{
  "mcpServers": {
    "contexin": {
      "command": "node",
      "args": ["/absolute/path/to/postmind-mcp-server/dist/index.js"],
      "env": {
        "POSTMIND_DATA": "/absolute/path/to/saved-posts.json"
      }
    }
  }
}