CNAP is built for agentic operations. Instead of clicking through dashboards, your AI agents manage clusters, deploy products, and automate infrastructure through the same API that powers the platform — via the Model Context Protocol (MCP). Two MCP servers provide full access to documentation and the platform API, turning any MCP-compatible tool into a CNAP control plane.
Quick Setup
Connect your AI tool in under a minute. Pick your editor:
Claude Code
Cursor
VS Code
Claude Desktop
Other
claude mcp add --transport http cnap-docs https://docs.cnap.tech/mcp
claude mcp add --transport http cnap https://cnap.tech/mcp
Press Cmd+Shift+P → “Open MCP settings” → Add:{
"mcpServers": {
"cnap-docs": {
"url": "https://docs.cnap.tech/mcp"
},
"cnap": {
"url": "https://cnap.tech/mcp"
}
}
}
Create .vscode/mcp.json:{
"servers": {
"cnap-docs": {
"type": "http",
"url": "https://docs.cnap.tech/mcp"
},
"cnap": {
"type": "http",
"url": "https://cnap.tech/mcp"
}
}
}
Config file location:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json
- Windows:
%APPDATA%\Claude\claude_desktop_config.json
- Linux:
~/.config/claude/claude_desktop_config.json
claude_desktop_config.json
{
"mcpServers": {
"cnap-docs": {
"command": "npx",
"args": ["mcp-remote", "https://docs.cnap.tech/mcp"]
},
"cnap": {
"command": "npx",
"args": ["mcp-remote", "https://cnap.tech/mcp"]
}
}
}
Add the MCP server URLs to your tool’s configuration:{
"mcpServers": {
"cnap-docs": {
"url": "https://docs.cnap.tech/mcp"
},
"cnap": {
"url": "https://cnap.tech/mcp"
}
}
}
Configuration format may vary. Refer to your tool’s documentation.
| Server | URL | Auth | Use case |
|---|
cnap-docs | https://docs.cnap.tech/mcp | None | Search and query docs |
cnap | https://cnap.tech/mcp | OAuth 2.1 | Manage clusters, products, deployments |
You can set up just one server if you only need docs search or only need API access.
What You Can Do
Once connected, your AI agents can:
- Deploy and manage — create products, spin up clusters, check deployment health, and scale infrastructure through natural language
- Chain operations — the platform MCP uses Code Mode so agents compose multiple API calls in a single step, like listing workspaces then checking each cluster’s status
- Query documentation — get instant answers about CNAP features, APIs, and configuration
- Stay current — MCP servers return real-time data from the platform, not stale training knowledge
List all clusters in my workspace and show me which ones have pending deployments.
Two MCP Servers
Beyond MCP
CNAP’s AI integration goes beyond MCP servers:
- Contextual menu — every doc page offers copy-as-markdown, open-in-AI, and one-click editor setup
- llms.txt — industry-standard files (
llms.txt and llms-full.txt) for AI indexing
- Markdown access — append
.md to any doc URL for AI-friendly output (e.g. docs.cnap.tech/ai/index.md)
