question: "What type of agent are you building?"
header: "Agent Architecture"
multiSelect: false
options:
  - label: "Single autonomous agent"
    description: "One agent with tools (ReAct, plan-and-execute)"
  - label: "Multi-agent system"
    description: "Multiple agents collaborating/delegating"
  - label: "Agentic RAG"
    description: "Agent-driven document retrieval"
  - label: "MCP integration (Claude tools)"
    description: "Model Context Protocol for Claude Code"

question: "Which agent framework?"
header: "Framework"
multiSelect: false
options:
  - label: "LangChain"
    description: "Most mature, extensive tooling"
  - label: "LlamaIndex"
    description: "Best for data/document agents"
  - label: "Custom implementation"
    description: "Direct API calls to LLM"
  - label: "Claude MCP (native)"
    description: "Claude-native tool protocol"

question: "What tools/capabilities does the agent need?"
header: "Agent Capabilities"
multiSelect: true
options:
  - label: "Web search"
    description: "Search internet for information"
  - label: "Code execution"
    description: "Run Python/JS code safely"
  - label: "Database queries"
    description: "Query SQL/NoSQL databases"
  - label: "API calls"
    description: "Call external REST/GraphQL APIs"
  - label: "File operations"
    description: "Read/write files, search codebase"

Input → Thought → Action → Observation → Thought → ... → Answer

Input → Plan → Execute Step 1 → Execute Step 2 → ... → Synthesize

Input → Extended Thinking → Plan → Execute → Answer

from langchain.agents import create_react_agent, AgentExecutor
from langchain_openai import ChatOpenAI
from langchain.tools import Tool

tools = [
    Tool(
        name="Calculator",
        func=lambda x: eval(x),
        description="Math calculator"
    )
]

llm = ChatOpenAI(model="gpt-4o")
agent = create_react_agent(llm, tools, prompt)
executor = AgentExecutor(agent=agent, tools=tools)

result = executor.invoke({"input": "What is 25 * 17?"})

from langchain.agents import initialize_agent, Tool
from langchain_openai import ChatOpenAI

# Define specialized agents
researcher = ChatOpenAI(model="gpt-4o")
writer = ChatOpenAI(model="gpt-4o")

# Orchestrator delegates tasks
orchestrator = initialize_agent(
    tools=[
        Tool(name="research", func=research_agent),
        Tool(name="write", func=writer_agent)
    ],
    llm=ChatOpenAI(model="gpt-4o"),
    agent="zero-shot-react-description"
)

result = orchestrator.invoke("Research AI trends and write a summary")

import anthropic

client = anthropic.Anthropic()

response = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=1024,
    tools=[{
        "name": "get_weather",
        "description": "Get weather data",
        "input_schema": {
            "type": "object",
            "properties": {
                "location": {"type": "string"}
            }
        }
    }],
    messages=[{"role": "user", "content": "Weather in NYC?"}]
)

from llama_index.agent import ReActAgent
from llama_index.llms import OpenAI
from llama_index.tools import QueryEngineTool

llm = OpenAI(model="gpt-4o")

tools = [
    QueryEngineTool.from_defaults(
        query_engine=query_engine,
        name="docs",
        description="Documentation search"
    )
]

agent = ReActAgent.from_tools(tools, llm=llm)
response = agent.chat("How do I use embeddings?")