Skip to main content

​
FastMCP Provider

FastMCP is a provider that enables direct tool execution through the Model Context Protocol (MCP). It provides a streamlined way to execute tools without complex workflow orchestration.

​
When to Use FastMCP

Use FastMCP when you:
  • Have MCP-compatible tools you want to execute
  • Need direct tool execution without AI workflow generation
  • Want to leverage the standardized MCP protocol
  • Are building integrations with existing MCP servers

​
Quick Start

​
1. Install Kubiya CLI

# Install Kubiya CLI
curl -fsSL https://raw.githubusercontent.com/kubiyabot/cli/main/install.sh | bash

# Set your API key
export KUBIYA_API_KEY="your-api-key"

​
2. Start MCP Server

# Start the Kubiya MCP server
kubiya mcp serve

​
3. Connect to MCP Server

# Test connection
kubiya mcp serve --test

# Or use with Claude Desktop
# Add to ~/.config/Claude/claude_desktop_config.json:
{
  "mcpServers": {
    "kubiya": {
      "command": "kubiya",
      "args": ["mcp", "serve"]
    }
  }
}

​
Available MCP Tools

The Kubiya MCP server provides 21+ enterprise-grade tools:

​
Core Execution Tools

execute_tool

Execute any containerized tool
kubectl get pods

execute_whitelisted_tool

Execute pre-approved tools
helm list

create_on_demand_tool

Build custom tools on-the-fly
python analyze_logs.py

execute_workflow

Run complete workflows
deploy-to-staging

​
Platform Management Tools

list_runners

List execution infrastructure
Show available runners

check_runner_health

Monitor system health
Health check all runners

find_available_runner

Auto-select optimal runners
Find best runner for task

list_agents

Discover AI agents
Show all agents

​
Knowledge & Security Tools

search_kb

Search organizational knowledge
Find documentation

list_kb

Browse documentation
Show knowledge base

list_secrets

View available credentials
Show secrets

list_integrations

See available integrations
Show integrations

​
Configuration

​
Basic Configuration

{
  "enable_runners": true,
  "allow_platform_apis": false,
  "enable_opa_policies": false,
  "verbose_logging": false
}

​
Enterprise Configuration

{
  "enable_runners": true,
  "allow_platform_apis": true,
  "enable_opa_policies": true,
  "verbose_logging": true,
  "whitelisted_tools": [
    {
      "name": "kubectl",
      "description": "Kubernetes CLI tool",
      "type": "docker",
      "image": "kubiya/kubectl-light:latest"
    }
  ]
}

​
Environment Variables

# Core configuration
export KUBIYA_API_KEY="your-api-key"
export KUBIYA_MCP_ENABLE_RUNNERS=true
export KUBIYA_MCP_ALLOW_PLATFORM_APIS=false

# Security
export KUBIYA_OPA_ENFORCE=true
export KUBIYA_MCP_REQUIRE_AUTH=true

# Logging
export LOG_LEVEL=INFO
export KUBIYA_MCP_VERBOSE=true

​
Usage Examples

​
Basic Tool Execution

# Ask your AI assistant:
"Use Kubiya to check the health of our Kubernetes cluster"

# The MCP server will execute:
# - kubectl get nodes
# - kubectl get pods --all-namespaces
# - kubectl top nodes

​
Infrastructure Operations

# Deploy application
"Deploy version 2.1.0 to staging using Kubiya"

# Database backup
"Create a backup of the production database"

# Log analysis
"Analyze the application logs for errors in the last hour"

​
Custom Tool Creation

# Create custom monitoring tool
"Create a tool that monitors disk space and sends alerts"

# Data processing
"Build a tool that processes CSV files and generates reports"

​
Integration Patterns

​
Claude Desktop Integration

{
  "mcpServers": {
    "kubiya": {
      "command": "kubiya",
      "args": ["mcp", "serve"],
      "env": {
        "KUBIYA_API_KEY": "your-api-key"
      }
    }
  }
}

​
Cursor IDE Integration

{
  "mcp.servers": {
    "kubiya": {
      "command": "kubiya",
      "args": ["mcp", "serve"],
      "env": {
        "KUBIYA_API_KEY": "your-api-key"
      }
    }
  }
}

​
Direct MCP Protocol

import asyncio
from mcp import ClientSession, StdioServerParameters
from mcp.client.stdio import StdioServerTransport

async def use_kubiya_mcp():
    server_params = StdioServerParameters(
        command="kubiya",
        args=["mcp", "serve"],
        env={"KUBIYA_API_KEY": "your-key"}
    )
    
    async with StdioServerTransport(server_params) as (read, write):
        async with ClientSession(read, write) as session:
            await session.initialize()
            
            # Execute a tool
            result = await session.call_tool(
                "execute_tool",
                {
                    "tool_name": "kubectl",
                    "args": {"command": "get pods"},
                    "runner": "auto"
                }
            )
            print(result.content)

asyncio.run(use_kubiya_mcp())

​
Security Features

​
OPA Policy Enforcement

# Enable policy enforcement
kubiya mcp serve --enable-opa-policies

​
Tool Whitelisting

{
  "whitelisted_tools": [
    {
      "name": "kubectl",
      "description": "Kubernetes commands",
      "type": "docker",
      "image": "kubiya/kubectl-light:latest"
    }
  ]
}

​
Authentication

# Require authentication
kubiya mcp serve --require-auth --session-timeout 3600

​
Monitoring and Logging

​
Debug Mode

# Enable verbose logging
kubiya mcp serve --verbose

# Or via environment
export LOG_LEVEL=DEBUG
kubiya mcp serve

​
Health Checks

# Test MCP server health
kubiya mcp serve --test

# Check runner health
kubiya runner health

​
Best Practices

  1. Security First
    • Always use API keys via environment variables
    • Enable OPA policies for production
    • Use tool whitelisting for restricted environments
  2. Performance
    • Use appropriate runners for different workloads
    • Monitor resource usage
    • Set reasonable timeouts
  3. Reliability
    • Implement proper error handling
    • Use health checks
    • Monitor execution logs

​
Troubleshooting

​
Common Issues

  1. Connection Failed 
    # Check MCP server status
kubiya mcp serve --test

# Verify API key
echo $KUBIYA_API_KEY
  2. Tool Execution Failed 
    # Check runner availability
kubiya runner list

# Test with simple command
kubiya tool execute --name "echo" --args "hello world"
  3. Permission Denied 
    # Check tool whitelist
kubiya mcp serve --list-tools

# Verify API key permissions
kubiya auth verify

​
Next Steps