Getting started

Set up jobContext with Claude Desktop

Five steps from zero to a fully connected AI memory layer for your job search. Takes about five minutes.

1

Install Python 3

Open Terminal and check if Python is already installed:

python3 --version

If you see "command not found", download Python from python.org/downloads and install it.

2

Clone the repo

git clone https://github.com/JustLikeFrank3/jobContextMCP
cd jobContextMCP
python3 -m venv .venv
.venv/bin/pip install -r requirements.txt
3

Create your config file

cp config.example.json config.json

Open config.json in any text editor. Fill in your name, email, phone, and LinkedIn. The openai_api_key field can be left blank for now — it is only needed for RAG search features, not tone samples or outreach.

4

Wire it to Claude Desktop

Make sure Claude Desktop is installed, then open this file in a text editor (create it if it does not exist):

Mac: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

Paste this, replacing the paths with wherever you cloned the repo:

{
  "mcpServers": {
    "jobContextMCP": {
      "command": "/absolute/path/to/jobContextMCP/.venv/bin/python3",
      "args": ["/absolute/path/to/jobContextMCP/server.py"],
      "cwd": "/absolute/path/to/jobContextMCP"
    }
  }
}

Example on Mac: /Users/yourname/jobContextMCP/.venv/bin/python3

5

Bootstrap your workspace

Restart Claude Desktop. Then in a Claude chat, say:

"Run setup_workspace and walk me through it."

The tool will ask for your info and create all your data files from scratch.

+

Add tone samples

Ask Claude to log_tone_sample from any message you write. After a few samples, get_tone_profile will reflect your voice and Claude can draft outreach in your register.

Note on work computers: IT restrictions may block running local scripts or modifying app config files. If you hit permission walls, Docker is the cleaner option. Check with IT first if unsure.

See also

Ready to connect?

Sign in to the hosted dashboard or run it locally.