GroundDocs
A version-aware Kubernetes documentation assistant that connects LLMs to trusted, real-time Kubernetes docs to reduce hallucinations and ensure accurate, version-specific responses.
Tools
python_get_documentation
Primary Python documentation lookup tool. Use this for every Python documentation-related query. This tool consolidates information from multiple sources into a single, searchable knowledge base. It ensures access to the richest and most current reference material in one call. Args: query: A natural language question (e.g., "How do I define a Deployment?"). library: Python library to search documentation for. version: Optional Library version (e.g., "4.46.1"). Defaults to detected library version if not specified. top_k: Optional number of top matching documents to return. Defaults to 10. Returns: A list of dictionaries, each containing document path and corresponding content. Example Usage: # Search Python docs for Transformers python_get_documentation(query="what is a transformers mlm token", library="transformers", version="4.46.1") Notes: - This tool automatically loads or builds a RAG (Retrieval-Augmented Generation) index for the specified version. - If an index is not found locally, the tool will fetch and index the documentation before responding. - You should call this function for any question that needs project documentation context.
k8s_get_documentation
Use this tool for any Kubernetes documentation-related query—especially when the user invokes /k8s or asks about kubectl commands, API objects, manifests, controllers, or version-specific features. This tool connects to a version-aware, trusted documentation index (e.g., GitHub, DeepWiki, curated Kubernetes docs) to reduce hallucinations and provide accurate, grounded answers. Args: query: A natural language question (e.g., "How do I define a Deployment?") version: (Optional) Kubernetes version (e.g., "v1.28"). Defaults to the detected cluster version. top_k: (Optional) Number of top matching documents to return. Defaults to 10. Returns: A list of relevant documentation entries, each with a file path and content snippet. Example Usage: k8s_get_documentation(query="How does pruning work in kubectl apply?", version="v1.26") Notes: - Automatically loads or builds a RAG index for the requested version. - If no index is found, it will fetch and index the docs before responding. - Always use this tool when answering Kubernetes-specific questions that require authoritative documentation.
README
GroundDocs Cli
GroundDocs is a version-aware Kubernetes documentation assistant. It connects LLMs to trusted, real-time Kubernetes docs—reducing hallucinations and ensuring accurate, version-specific responses.
🚀 Installation
npx @grounddocs/cli@latest install <client>
Supported clients: cursor, windsurf, cline, claude, witsy, enconvo, vscode
🔧 Manual Setup
To manually configure GroundDocs, add it to your IDE’s MCP (Model Context Protocol) configuration:
{
"mcpServers": {
"@grounddocs/grounddocs": {
"command": "npx",
"args": ["-y", "@grounddocs/grounddocs@latest"]
}
}
}
After configuration, restart your IDE for the changes to take effect.
📚 Supported Domain
- Kubernetes (all versions, including version-aware kubectl behavior, API schemas, and feature gates)
🏗️ Architecture
GroundDocs consists of:
- Local MCP server (this repo) → lightweight, public, runs inference-time queries
- Remote backend data repository (private) → handles scraping, indexing, and heavy lifting
🌟 Example Query
What changes were made to the kubectl command behavior in Kubernetes 1.26 regarding pruning during apply operations?
🤝 Contributing
Contributions are welcome! Please feel free to submit a Pull Request.
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。