What problem does Holster solve?
Holster replaces manual text editing of Claude’s config file when managing MCP server entries, allowing Claude to handle the task directly.
@kimjune01
基本情報
MCP server that will edit your claude configs
以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。
{
"mcpServers": {
"mcp-holster": {
"command": "python",
"args": [
"-m",
"venv",
".venv"
]
}
}
}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。
Holster is a Python-based MCP server that manages other MCP servers in Claude’s desktop configuration file directly from within Claude, eliminating the need for manual text editor modifications. It is designed for users who frequently add or remove MCP servers and want to do so through Claude’s interface.
Install Holster by cloning the repository, creating a virtual environment, and installing dependencies with uv. Then add the Holster server entry to your claude_desktop_config.json (default location: ~/Library/Application Support/Claude/claude_desktop_config.json). After restarting Claude, use the provided tools (explain_holster, create_server, list_servers, update_server_status, delete_servers, ping) to manage server configurations.
Holster replaces manual text editing of Claude’s config file when managing MCP server entries, allowing Claude to handle the task directly.
Holster requires Python, a virtual environment, and `uv` for dependency management. It runs as a local MCP server.
Holster reads and writes to `~/Library/Application Support/Claude/claude_desktop_config.json` (Claude’s configuration file). It does not manage servers outside that file.
No. Server discovery and downloading are explicitly out of scope. Holster only manages configurations already present in the config file.
The README does not specify a transport. It is configured as a local MCP server using the `uv` command and `holster.py` script.
Painless E2E Automation for Mobile and Web

See and control a real Windows PC you own — from any MCP client, locally or remotely. Observe (UIA + screenshots), click/type/drag/scroll, launch apps, owner Live View. BYOH: your machine, your key.
AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

Handl speaks the Model Context Protocol. Your whole billing operation — projects, invoices, clients, reports, approvals — as tools any AI assistant can call, inside your autonomy rules.
不止于 CSS
コメント