Skip to content

Configuration

Configure Notify-MCP to work with your AI assistant and team.

Claude Desktop / Claude Code

Add Notify-MCP to your Claude configuration file:

File Location: - macOS/Linux: ~/.config/Claude/claude_desktop_config.json - Windows: %APPDATA%\Claude\claude_desktop_config.json

Configuration:

{
  "mcpServers": {
    "notify-mcp": {
      "command": "uv",
      "args": ["run", "python", "-m", "notify_mcp"],
      "cwd": "/absolute/path/to/notify-mcp/packages/community",
      "env": {
        "NOTIFY_MCP_STORAGE_TYPE": "sqlite",
        "NOTIFY_MCP_SQLITE_PATH": "~/.notify-mcp/storage.db",
        "NOTIFY_MCP_MAX_HISTORY": "1000"
      }
    }
  }
}

Restart Claude Desktop after saving the configuration.

Storage Configuration

In-Memory Storage (Default)

Fast but not persistent across restarts. Good for individual use:

{
  "env": {
    "NOTIFY_MCP_STORAGE_TYPE": "memory"
  }
}

Persistent storage that enables team collaboration:

{
  "env": {
    "NOTIFY_MCP_STORAGE_TYPE": "sqlite",
    "NOTIFY_MCP_SQLITE_PATH": "~/.notify-mcp/storage.db",
    "NOTIFY_MCP_MAX_HISTORY": "1000"
  }
}

For team collaboration, use a shared network location:

{
  "env": {
    "NOTIFY_MCP_STORAGE_TYPE": "sqlite",
    "NOTIFY_MCP_SQLITE_PATH": "/shared/team/notify-mcp.db"
  }
}

All team members pointing to the same database file will share channels and notifications!

Detailed Storage Configuration

Configuration Reference

Variable Options Default Description
NOTIFY_MCP_STORAGE_TYPE memory, sqlite memory Storage backend type
NOTIFY_MCP_SQLITE_PATH file path ~/.notify-mcp/storage.db SQLite database path
NOTIFY_MCP_MAX_HISTORY integer 1000 Max notifications per channel

ChatGPT Configuration

ChatGPT MCP support is in development. Use an MCP bridge for now:

# 1. Run notify-mcp server
uv run python -m notify_mcp

# 2. Use MCP-to-HTTP bridge
mcp-bridge --stdio "uv run python -m notify_mcp" --http localhost:8080

# 3. Configure ChatGPT Custom GPT to use the bridge endpoint

Native ChatGPT MCP support expected in 2025.

Gemini Configuration

Gemini MCP integration is in development. Use the Gemini API with an MCP adapter:

from google.generativeai import GenerativeModel
from mcp import ClientSession, StdioServerParameters
from mcp.client.stdio import stdio_client

async def main():
    server_params = StdioServerParameters(
        command="uv",
        args=["run", "python", "-m", "notify_mcp"]
    )

    async with stdio_client(server_params) as (read, write):
        async with ClientSession(read, write) as session:
            await session.initialize()
            # Use Gemini with MCP tools

Full Gemini MCP integration expected in 2025.

Slash Commands (Claude Code)

Notify-MCP includes 6 convenient slash commands:

  • /notify-decision - Create architecture decision notification
  • /notify-alert - Create critical alert
  • /notify-status - Share status update
  • /notify-channels - List all channels
  • /notify-subscribe - Subscribe to a channel
  • /notify-recent - View recent notifications

These commands are automatically available in Claude Code once Notify-MCP is configured.

Next Steps

Troubleshooting

Server Not Starting

# Test manually
uv run python -m notify_mcp

# Check if uv is in PATH
which uv

# Use absolute path if needed
"command": "/full/path/to/uv"

Claude Doesn't See the Server

  1. Check the configuration file path is correct
  2. Ensure you restarted Claude Desktop
  3. Verify the cwd path points to the correct directory
  4. Check Claude logs for error messages

For more help, see the Troubleshooting Guide.