Process Mining MCP Server

Project Launch Instructions

IMPORTANT: The project was run on a Windows system.

Steps to follow

1. Install PostgreSQL, create a database and a table with any name.
   Access credentials (database name, user, password, table name, etc.) should be entered in the .env file.
   There is a sample .env.example file in the repository — use it as a template to fill in your actual .env file.
2. Create an account on Anthropic and add funds to it.
   Otherwise, I can share my API_KEY via email.
3. The repository contains example .bat launch files with the _example suffix.
   Create your own .bat files (without _example in the name), filling in the appropriate project paths.
4. Create a virtual environment and install the required libraries according to requirements.txt:

pip install -r requirements.txt

5. Download Claude Desktop.
   Go to File -> Settings -> Developer -> Edit Config.
   Paste the contents of the claude_desktop_config_example.json, file there, filling in the paths to your .bat files.

   Make sure to use the correct path format.

6. Extract the file BPI Challenge 2017.xes.gz into the datasets folder.
7. Run the data import script:

python import_data.py

The script should import the event log data into the database.

8. Restart Claude Desktop.
   After launching the tool, the tools should be visible below the input field (two sliders labeled "search and tools"). There you can enable or disable the available tools.

Optional: Configuring MCP in Cursor

If you are using Cursor with MCP tools:

1. Go to File -> Preferences -> Cursor Settings.
2. Go to MCP Tools tab.
3. Press New MCP Server button and paste previously filled claude_desktop_config.json file contents, adjusting the paths to your .bat files as needed.
4. Save the configuration. Your MCP tools should now be available inside Cursor.