Connect your assets to AI.
Markie exposes your local asset library via MCP, so Claude and other AI tools can search and retrieve your files.
What is MCP?
Model Context Protocol (MCP) is a standard for AI assistants to access external tools and data sources securely.
Markie's MCP server is available as an npm package and exposes your personal asset library to Claude and other AI tools. Search by query, tag, file type, or description — all from within your AI conversations.
Your files stay local. Your AI accesses them through Markie, which controls exactly what can be searched and retrieved.
Example usage.
“Find that blue gradient poster I saved last week”
“Find all assets tagged "logo"”
“Tell me more about that gradient poster”
“Get me the brand guidelines PDF”
“Show me a preview of that poster”
“What tags do I use most often?”
“What colors are in my library?”
“What did I save recently?”
“How big is my library?”
Available Tools
Use these MCP tools to interact with your Markie library programmatically. View full documentation →
search_assets
Full-text search with filters for type, tags, date range, and colors. Returns matching assets with metadata.
get_asset_detail
Get complete metadata for a specific asset including tags, colors, and source info.
get_asset_file
Retrieve the original file as base64. Maximum 10 MB.
get_asset_thumbnail
Retrieve a thumbnail image for visual preview of a specific asset.
list_tags
List all tags in your library, ranked by frequency.
list_colors
List all dominant colors across your library, ranked by frequency.
get_recent
Get the most recently saved assets in your library.
get_stats
Library statistics including asset counts, total size, and file type breakdown.
Quick Start
Claude Code
claude mcp add markie -- npx -y @vitgranen/markie-mcp
Claude Desktop — add to ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"markie": {
"command": "npx",
"args": ["-y", "@vitgranen/markie-mcp"]
}
}
}Other MCP Clients — use as stdio command
npx -y @vitgranen/markie-mcp