The MythOS MCP Server is an 📝npm package that connects your 📝MythOS knowledge library to any MCP-compatible AI client — 📝Claude Desktop, Cursor, Windsurf, VS Code + GitHub Copilot, and more. Search, read, create, and update memos, chat with your library using RAG, explore tags, and browse communities — all from your AI tool of choice. The package requires a MythOS Scholar or Oracle plan as an 📝API key is necessary.
How to Install
- Generate an API key in Settings > API.
- Add the following to your MCP client configuration (e.g., Claude Desktop at
~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"mythos": {
"command": "npx",
"args": ["-y", "mythos-mcp"],
"env": {
"MYTHOS_API_KEY": "mtk_your_key_here",
"MYTHOS_USERNAME": "your_username"
}
}
}
}- Restart your client for the MCP connection to take effect.
How to Use
Once connected, your AI assistant has access to 10 tools for interacting with your library:
- search_memos — find memos by keyword, tag, or visibility
- read_memo — read full content with optional section extraction
- create_memo — create new memos with title, content, and tags
- update_memo — update existing memo content and tags
- list_tags — see all your tags with usage counts
- chat_with_library — RAG-powered Q&A over your knowledge base
- delta_sync — check what changed since a given timestamp
- get_memo_index — lightweight overview of your entire library
- list_communities — discover public MythOS communities
- search_community_posts — search posts within a community
How it Works
Architecture: The package runs locally as a stdio MCP server, making fetch() calls directly to the MythOS API at mythos.one. No proxy, no local database — just a thin bridge between your AI client and your MythOS library. One dependency (@modelcontextprotocol/sdk), 7.5 kB package size.
