WEATHGARDS
Exposes MCP tools that enable remote LLMs to query local Docker containers, OS processes, and system services in real time.
README
WEATHGARDS
Local cross-OS MCP gateway for Docker containers, processes and services.
WEATHGARDS scans your machine for running Docker containers, OS processes, and system services, then exposes pre-written MCP tools over a secure local network endpoint so a remote LLM (Claude, GPT-4, etc.) can interrogate your environment in real time — without generating any code at runtime.
Prerequisites
Docker (optional but recommended)
WEATHGARDS works without Docker; container scanning is simply disabled when the daemon is unreachable.
| OS | Install |
|---|---|
| Linux | Install the Docker Engine daemon: sudo apt install docker.io (Debian/Ubuntu) or follow docs.docker.com. Add your user to the docker group: sudo usermod -aG docker $USER then log out and back in. |
| macOS | Install Docker Desktop for Mac. Start the app and wait for the whale icon in the menu bar. |
| Windows | Install Docker Desktop for Windows (requires WSL 2 or Hyper-V). Start Docker Desktop before running WEATHGARDS. |
Python 3.12+
# Linux / macOS
python3 --version # must be ≥ 3.12
# Windows (PowerShell)
python --version
Install uv — the package manager used throughout:
# Linux / macOS
curl -LsSf https://astral.sh/uv/install.sh | sh
# Windows (PowerShell)
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
Installation
# 1. Clone
git clone https://github.com/your-org/weathgards.git
cd weathgards
# 2. Install Python dependencies
uv sync
# 3. Configure environment
cp .env.example .env # Linux / macOS
copy .env.example .env # Windows (cmd)
# Edit .env — at minimum set WEATHGARDS_MCP_TOKEN
# 4. Build the frontend (one-time, requires Node 18+)
cd frontend && npm install && npm run build && cd ..
# 5. Start WEATHGARDS
uv run weathgards
Open http://127.0.0.1:8765 in your browser.
Per-OS Quick Reference
Linux
uv sync
cp .env.example .env
# nano .env → set WEATHGARDS_MCP_TOKEN
uv run weathgards
Docker socket at /var/run/docker.sock. If missing, container scanning is skipped automatically.
macOS
uv sync
cp .env.example .env
# open -e .env → set WEATHGARDS_MCP_TOKEN
uv run weathgards
Docker Desktop must be running before WEATHGARDS starts. The socket is detected automatically under both /var/run/docker.sock and the Docker Desktop path.
Windows (PowerShell)
uv sync
Copy-Item .env.example .env
# notepad .env → set WEATHGARDS_MCP_TOKEN
uv run weathgards
Docker Desktop must be running. The named pipe \\.\pipe\docker_engine is checked automatically. No administrator privileges required.
Connect a Remote LLM
Once WEATHGARDS is running with network exposure enabled, your LLM client needs three pieces of information: the endpoint URL, the Bearer token, and the MCP client config snippet.
Step 1 — Enable network exposure
In the Serveur tab of the cockpit UI:
- Toggle "Exposer sur le réseau local"
- Click "Démarrer le serveur MCP"
Or configure it persistently in .env before starting:
WEATHGARDS_MCP_TOKEN=your-secret-token-here
WEATHGARDS_HOST=0.0.0.0
Step 2 — Retrieve the endpoint and token
The Informations de connexion section shows the live values:
Endpoint MCP : http://192.168.1.42:9766/mcp
Token : wg-xxxxxxxxxxxxxxxxxxxx
Step 3 — Configure your MCP client
Paste the generated JSON snippet (available via the "Copier" button) into your Claude Desktop / MCP client config:
{
"mcpServers": {
"weathgards": {
"url": "http://192.168.1.42:9766/mcp",
"headers": {
"Authorization": "Bearer wg-xxxxxxxxxxxxxxxxxxxx"
}
}
}
}
Claude Desktop config path:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json- Windows:
%APPDATA%\Claude\claude_desktop_config.json- Linux:
~/.config/Claude/claude_desktop_config.json
⚠ Security Warning
Network exposure requires authentication — no exceptions.
Binding the MCP server to 0.0.0.0 makes it reachable by every device on your LAN. WEATHGARDS enforces a Bearer token on every /mcp request and refuses to start in network mode without a configured token.
Recommended hardening:
- Set
WEATHGARDS_MCP_TOKENto a long random string (≥ 32 characters):python3 -c "import secrets; print('wg-' + secrets.token_hex(32))" - Restrict access to a specific subnet via
WEATHGARDS_ALLOWED_CLIENTS=192.168.1.0/24in.env. - Never commit your
.envfile — it is already in.gitignore. - Do not expose WEATHGARDS on a public network. It is designed for a trusted local subnet only.
Development
# Install with dev extras
uv sync --extra dev
# Run with auto-reload
uv run weathgards --reload
# Lint
uv run ruff check src tests
# Format
uv run ruff format src tests
# Type check
uv run mypy src
# Tests (with coverage)
uv run pytest
# Tests — fast, no coverage
uv run pytest --no-cov
# Smoke tests only
uv run pytest tests/test_smoke.py --no-cov -v
# Build frontend
cd frontend && npm run build
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。