PyProc Logo PyProc v0.3.3

MCP Server Configuration

The Model Context Protocol (MCP) server allows LLM clients to communicate directly with PyProc's tools. PyProc supports two interface execution modes: stdio (Standard Input/Output) and http (SSE-based Server-Sent Events).

Interface Types
  • stdio: The client spawns the MCP server as a sub-process and communicates via standard input/output. This is optimal for local applications like Claude Desktop, Claude Code, and Cursor.
  • http: The MCP server runs as a standalone HTTP server using Server-Sent Events (SSE). This is useful for remote setups, virtualized workspaces, or integrations where a permanent connection is preferred.

Generating Configuration Profiles

To easily construct schema-compliant configuration lines for your AI environment, use the built-in generator options. This command will print the JSON configuration block directly onto your shell terminal.

1. Stdio Interface Configuration

Generate the configuration profile for standard process client integrations:

bash 
pyproc mcp --generate-config

This command prints the following JSON configuration to stdout:

json (generated config) 
{
  "mcpServers": {
    "pyproc": {
      "command": "/home/user/projects/pyproc/.venv/bin/pyproc-mcp",
      "env": {
        "PYPROC_TIMEOUT": "30",
        "PYPROC_RATE_LIMIT_DELAY": "1.0",
        "PYPROC_LOG_LEVEL": "INFO",
        "PYPROC_SSL_VERIFY": "0",
        "PYPROC_MCP_WORKERS": "4"
      }
    }
  }
}

2. HTTP Interface Configuration

Generate the configuration profile for HTTP-SSE connections (e.g. running the server on port 9090):

bash 
pyproc mcp --interface http --port 9090 --generate-config

This command prints the following JSON configuration to stdout:

json (generated config) 
{
  "mcpServers": {
    "pyproc": {
      "type": "streamableHttp",
      "url": "http://localhost:9090/"
    }
  }
}