🤖 Agent Runtimes

Agent Runtimes is a unified platform for deploying, managing, and interacting with AI agents across multiple protocols and frameworks. It provides both a Python server for hosting agents and React components for seamless integration into web and desktop applications.

🎯 What is Agent Runtimes?

Agent Runtimes solves the complexity of deploying AI agents by providing:

1. Protocol Abstraction: One agent, multiple protocols - deploy your agent once and access it through ACP, Vercel AI SDK, AG-UI, MCP-UI, or A2A without changing your code.

2. Framework Flexibility: Write agents using your preferred framework (Pydantic AI, LangChain, Jupyter AI) while maintaining a consistent API.

3. Cloud Runtime Management: Built-in integration with Datalayer Cloud Runtimes for launching and managing compute resources with Zustand-based state management.

4. UI Components: Pre-built React components (ChatBase, ChatSidebar, ChatFloating) that connect to agents and execute tools directly in the browser.

5. Tool Ecosystem: Seamless integration with MCP (Model Context Protocol) tools, custom tools, and built-in utilities for Jupyter notebooks and Lexical documents.

🌟 Features

Multi-Protocol Support

• ACP (Agent Client Protocol): WebSocket-based standard protocol
• Vercel AI SDK: Compatible with Vercel's AI SDK for React/Next.js
• AG-UI: Lightweight web interface (Pydantic AI native)
• MCP-UI: Interactive UI resources protocol with React/Web Components
• A2A: Agent-to-agent communication

Multi-Agent Support

• Pydantic AI: Type-safe agents (fully implemented)
• LangChain: Complex workflows (adapter ready)
• Jupyter AI: Notebook integration (adapter ready)

Built-in Features

• 🔌 Flexible Architecture: Easy to add new agents and protocols
• 🛠️ Tool Support: MCP, custom tools, built-in utilities
• 📊 Observability: OpenTelemetry integration
• 💾 Persistence: DBOS support for durable execution
• 🔒 Context Optimization: LLM context management

🏗️ Architecture

Agent Runtimes consists of three main components:

1. Python Server (agent_runtimes/)

The backend server that hosts AI agents and handles protocol routing:

• Agent Adapters: Unified interface for Pydantic AI, LangChain, and Jupyter AI
• Protocol Adapters: Convert between different agent protocols (ACP, AG-UI, Vercel AI, etc.)
• FastAPI Server: High-performance async server with automatic API documentation
• Tool Registry: Manages and executes tools from various sources (MCP, custom, built-in)

2. React Components (src/components/)

Pre-built UI components for interacting with agents:

• ChatBase: Core chat interface with customizable styling
• ChatSidebar: Collapsible sidebar for agent interactions
• ChatFloating: Floating chat widget for non-intrusive agent access
• All components support: Frontend tool execution, markdown rendering, code highlighting, and real-time streaming

3. Runtime Management (src/runtime/)

Cloud runtime lifecycle management with Zustand store:

• Launch & Connect: Create new cloud runtimes or connect to existing ones
• Agent Creation: Automatically create and configure agents on runtimes
• State Management: Track runtime status, agent connections, and errors
• Hooks: React hooks for easy integration (useAgentRuntime, useRuntimeStore)

🚀 Use Cases

Notebook AI Assistant

Add an AI agent to Jupyter notebooks that can:

• Execute cells, insert code, and modify notebook content
• Explain code and data analysis
• Debug errors and suggest improvements

import { NotebookAgentsRuntime } from '@datalayer/agent-runtimes';

<NotebookAgentsRuntime
  notebookId={notebookId}
  environmentName="python-simple"
  runtimeName={runtimeName}
  serviceManager={serviceManager}
/>

Document Editor AI

Integrate AI into Lexical-based document editors:

• Insert headings, lists, code blocks, and formatted text
• Summarize content and proofread text
• Generate ideas and help with writing

import { DocumentAgentRuntime } from '@datalayer/agent-runtimes';

<DocumentAgentRuntime
  documentId={documentId}
  environmentName="python-simple"
  runtimeName={runtimeName}
  serviceManager={serviceManager}
/>

Custom Agent Deployment

Deploy your own Pydantic AI agent with custom tools:

from agent_runtimes import AgentRuntimesApp
from pydantic_ai import Agent

# Create your agent
agent = Agent(
    model='anthropic:claude-sonnet-4-5',
    system_prompt='You are a helpful assistant.',
)

# Launch the server
app = AgentRuntimesApp()
app.add_agent(agent, name='my-agent', transport='ag-ui')
app.run(port=8000)

🔧 Key Concepts

Protocols

Agent Runtimes supports multiple protocols for agent communication:

• AG-UI: Lightweight protocol for web UIs (POST-based, Pydantic AI native)
• ACP: WebSocket-based Agent Client Protocol for real-time interaction
• Vercel AI SDK: Compatible with Vercel's AI SDK streaming
• MCP-UI: Model Context Protocol with UI resources
• A2A: Agent-to-agent communication protocol

Tools

Tools extend agent capabilities by allowing them to perform actions:

• Frontend Tools: Execute in the browser (notebook editing, document manipulation)
• MCP Tools: Tools from Model Context Protocol servers
• Custom Tools: Your own Python functions decorated with tool metadata
• Built-in Tools: File operations, web search, code execution
• Code Mode: Tool discovery includes output_schema and input_examples for reliable calls; code execution returns stdout/stderr and a summarized result.

Runtime Management

Cloud runtimes provide compute resources for agents:

import { useAgentRuntime } from '@datalayer/agent-runtimes/lib/runtime';

const { isReady, endpoint, tools, launchRuntime } = useAgentRuntime({
  autoCreateAgent: true,
  agentConfig: {
    model: 'anthropic:claude-sonnet-4-5',
    systemPrompt: 'You are a helpful AI assistant.',
  },
});

// Launch a new runtime
await launchRuntime({
  environmentName: 'python-simple',
  creditsLimit: 100,
  type: 'notebook',
});