Full Stack Agents

Build AI-powered applications using Kubiya workflows with different providers

Full Stack Agents

Full Stack Agents combine Kubiya’s deterministic workflow engine with AI providers to create powerful automation solutions. You can choose from different providers based on your needs.

Available Providers

ADK Provider

Natural language to workflow generation

Use ADK when you want to:

• Generate workflows from plain English
• Let AI plan and execute complex tasks
• Stream real-time progress to users

adk = get_provider("adk")
result = await adk.compose("Deploy my app with health checks")

FastMCP Provider

Direct tool execution via MCP protocol

Use FastMCP when you want to:

• Connect existing MCP-compatible tools
• Execute specific tools directly
• Build on standardized protocols

mcp = get_provider("fastmcp")
result = await mcp.execute_tool("database_query", params)

How It Works

Both providers integrate with Kubiya workflows:

Quick Example: Using ADK

from kubiya_workflow_sdk.providers import get_provider

# 1. Get the ADK provider
adk = get_provider("adk", api_key="YOUR_KEY")

# 2. Generate and execute a workflow
result = await adk.compose(
    task="Create a Python script that analyzes CSV files",
    mode="act",  # Execute immediately
    stream=True  # Stream progress
)

# 3. Handle streaming results
async for event in result:
    print(event)  # Real-time updates

Quick Example: Using FastMCP

from kubiya_workflow_sdk.providers import get_provider

# 1. Get the FastMCP provider
mcp = get_provider("fastmcp", server_path="/path/to/mcp/server")

# 2. Execute a tool directly
result = await mcp.execute_tool(
    tool_name="file_operations",
    arguments={"action": "read", "path": "/data/report.csv"}
)

print(result)  # Tool execution result

Building Full Stack Applications

1. Choose Your Provider

• ADK: For natural language interfaces and complex automation
• FastMCP: For direct tool integration and MCP compatibility

2. Add Frontend

Use the Vercel AI SDK or any frontend framework to create user interfaces.

3. Deploy

Run agent servers locally or deploy to production with Helm charts.

Integration with Existing Frameworks

You can also integrate Kubiya with existing AI frameworks:

LangGraph

Use LangGraph agents within Kubiya workflows

Other Frameworks

Integrate LangChain, CrewAI, and more

Next Steps

1. Pick a Provider

   Choose between ADK for AI generation or FastMCP for tool execution

2. Build Your Agent

   Follow the provider documentation to create your agent server

3. Add UI (Optional)

   Create a frontend for user interaction

4. Deploy

   Run locally or deploy to production