Khwan is a Model Context Protocol (MCP) server: Durable AI memory: recall distilled facts in a later session or subagent, and remember new ones. In practice that means any MCP-compatible AI assistant can call Khwan's tools directly — the model decides when to use them in a conversation or agent run.
You can run Khwan two ways: connect to the hosted endpoint at https://mcp.khwan.ai/mcp/{core}/{user} with no local install, or run it on your own machine via the Python package khwan-mcp (via uvx). Configuration happens through environment variables — you'll need KHWAN_API_KEY. The current release is v0.3.12, published under the MIT license, with source at khwanlabs/khwan-mcp on GitHub.
Khwan is listed under Memory on mcp.site and works with any MCP client — Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, Zed and the rest of the ecosystem — using the install snippets below. If you maintain Khwan for khwan.ai, claim this listing to verify ownership, earn the Verified badge, and keep the details current.