What is a Desktop Extension (.dxt)?
A `.dxt` file is a zip archive containing a local MCP server and a `manifest.json`. It is similar to Chrome `.crx` or VS Code `.vsix` files, enabling single-click installation of MCP servers.
@anthropics
基本情報
概要はまだありません
ツール一覧はまだありません
この掲載には表示できるツール一覧がまだありません。サーバーにツールがないという意味ではありません。対応機能はプロジェクトのドキュメントをご確認ください。
ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。
Rehe implements the Desktop Extensions (DXT) format, a specification for packaging local MCP servers into single-file .dxt archives that can be installed with one click. It includes a CLI tool for creating extensions, the manifest schema, and the loader used by Claude for macOS and Windows.
Install the CLI globally via npm install -g @anthropic-ai/dxt, then in your MCP server’s folder run dxt init to create a manifest.json and dxt pack to produce the .dxt file. Open the resulting file with an app that supports DXT (e.g., Claude for macOS or Windows) to install.
.dxt zip archivemanifest.json for metadata and configurationdxt init and dxt pack) for extension creationA `.dxt` file is a zip archive containing a local MCP server and a `manifest.json`. It is similar to Chrome `.crx` or VS Code `.vsix` files, enabling single-click installation of MCP servers.
The CLI requires Node.js and is installed globally via `npm install -g @anthropic-ai/dxt`. No other runtime dependencies are needed beyond those of the MCP server being packaged.
The extension is a self-contained zip file. When installed, the app (e.g., Claude) loads the server from the unpacked archive on the local filesystem.
For Python, bundle packages in `server/lib/` or a full virtual environment in `server/venv/`. For Node.js, run `npm install --production` and bundle the `node_modules` directory. The manifest’s `server.entry_point` points to the main executable or file.
The MCP server communicates over stdio transport. Authentication is handled by the server itself; Rehe does not impose its own auth layer. The manifest can specify environment variables and parameters via `mcp_config.env`.

DeepMark helps teachers deliver rapid, consistent marking with meaningful feedback for every student — in a fraction of the time. What once took a week, now takes one free period.
Open-source MCP to use Blender with any LLM
Hindsight: Agent Memory That Learns

Model Context Protocol Servers

Yuppsy MCP: build and publish real estate websites from your AI assistant. Yuppsy lets your AI assistant create live websites for real estate agents and individual property listings. You don't need a designer or any code
コメント