Memories Wrapped

Memories Wrapped MCP Server creates a concise summary of your Apple Photos memories by running a dedicated year-wrapped process through the MCP framework. It is useful when you want an automated, organized recap of your photo memories across years, accessible via an MCP client workflow.

How to use

You interact with Memories Wrapped by requesting a year-by-year memory summary through your MCP client. After setting up the server, you can trigger the year-wrapped analysis for your Apple Photos collection, and the server will return a structured summary you can review, save, or share.

How to install

Prerequisites and initial setup are straightforward. You need a host environment with a compatible MCP runtime available (uv) and a terminal to run commands.

# Step 1: Install the MCP runtime if you have not yet
# The runtime command is typically provided by uvx in your environment
# Ensure uv is installed and accessible

# Step 2: Clone the project repository (example path shown; replace with your actual repo path)
# git clone git@github.com:niyabits/memories-wrapped-mcp-server.git

# Step 3: Sync dependencies using the MCP runtime integration
uv sync

# Step 4: Configure the MCP Server with the runtime command and script
# Copy the configuration below and tailor <username> and <repository-path> to your setup
# This is the standard MCP JSON configuration snippet

This is the MCP configuration you will place in your settings to run the year-wrapped process. Replace the placeholder values for your environment before using it.

{
  "mcpServers": {
    "year-wrapped": {
        "command": "/Users/<username>/.local/bin/uv",
        "args": [
          "--directory",
          "<repository-path>/memories-wrapped-mcp-server",
          "run",
          "year_wrapped.py"
        ]
      }
    },
}

Make sure to replace the <username> and <repository-path> with your actual username and the location where you cloned the project. You can determine the actual runtime path for uv by running: which uv. The <repository-path> is where you cloned the project.

Finally, open Claude and request your year-wrapped output to start generating your memory recap.

Additional sections

Notes and best practices: keep your repository path up to date if you relocate the project, and ensure the year_wrapped.py script has the correct permissions to run under your user account. If you encounter permission issues, verify that the uv binary and the script have execute permissions for your user.

Security and access: treat your local MCP server configuration as you would any local development credential. Do not expose the MCP server to untrusted networks without proper access control.