LangGraph Browser Example

This example demonstrates using AgentCore Browser with LangChain/LangGraph for web automation and research tasks.

Features

LangChain Integration - Uses langchain_aws.tools.create_browser_toolkit
React Agent - LangGraph's create_react_agent for tool orchestration
Full Browser Control - Navigate, click, type, extract, screenshot

Project Structure

langgraph-browser/
├── serverless.yml      # Serverless configuration
├── agent.py            # LangGraph agent with browser toolkit
├── pyproject.toml      # Python dependencies
├── Dockerfile          # Container configuration
├── test-invoke.py      # Test script
└── README.md           # This file

Quick Start

1. Deploy

serverless deploy

2. Note the Runtime Endpoint URL

After deployment, note the runtime endpoint URL from the output (the ARN is embedded in the URL path):

ai:
  agents:
    browserAgent: https://bedrock-agentcore.us-east-1.amazonaws.com/runtimes/arn:aws:bedrock-agentcore:us-east-1:123456789012:runtime/xxx/invocations

3. Test

export RUNTIME_ARN="arn:aws:bedrock-agentcore:us-east-1:123456789012:runtime/xxx"
python test-invoke.py

How It Works

Browser Toolkit

The example uses langchain_aws.tools.create_browser_toolkit which provides these tools:

Tool	Description
`navigate_browser`	Navigate to a URL
`click_element`	Click on an element using CSS selector
`type_text`	Type text into an input field
`extract_text`	Extract all text content from the page
`extract_hyperlinks`	Extract all hyperlinks from the page
`get_elements`	Get elements matching a CSS selector
`current_webpage`	Get the current page URL and title
`navigate_back`	Go back to the previous page
`take_screenshot`	Take a screenshot of the page
`scroll_page`	Scroll the page in a direction
`wait_for_element`	Wait for an element to appear

LangGraph Agent

from langchain.chat_models import init_chat_model
from langchain_aws.tools import create_browser_toolkit
from langgraph.prebuilt import create_react_agent

# Create toolkit
toolkit, browser_tools = create_browser_toolkit(region="us-east-1")

# Initialize chat model (default: global.anthropic.claude-sonnet-5, override via MODEL_ID)
llm = init_chat_model(
    MODEL_ID,
    model_provider="bedrock_converse",
)

# Create agent with browser tools
agent = create_react_agent(
    model=llm,
    tools=browser_tools,
)

# Run with session isolation
config = {"configurable": {"thread_id": "session-123"}}
result = await agent.ainvoke(
    {"messages": [{"role": "user", "content": "Navigate to example.com"}]},
    config=config
)

Session Isolation

Each thread_id gets its own browser session, enabling concurrent usage:

# Each thread gets its own browser session
config_user1 = {"configurable": {"thread_id": "user-1"}}
config_user2 = {"configurable": {"thread_id": "user-2"}}

Example Prompts

Navigation:

Navigate to https://example.com and tell me the main heading

Extract Links:

Navigate to https://aws.amazon.com and extract the first 5 hyperlinks

Page Analysis:

Navigate to https://python.org and describe the main sections

Form Interaction:

Navigate to https://google.com, type "AWS Lambda" in the search box

Dependencies

langchain-aws - AWS integrations for LangChain
langgraph - Agent orchestration
playwright - Browser automation (required by browser toolkit)
beautifulsoup4 - HTML parsing

Reference

Cleanup

serverless remove

Next Steps

Gateway Example - Add custom Lambda tools
Memory Example - Add conversation persistence
Browser Documentation - Full configuration reference

LangGraph Browser Example

This example demonstrates using AgentCore Browser with LangChain/LangGraph for web automation and research tasks.

Features

LangChain Integration - Uses langchain_aws.tools.create_browser_toolkit
React Agent - LangGraph's create_react_agent for tool orchestration
Full Browser Control - Navigate, click, type, extract, screenshot

Project Structure

langgraph-browser/
├── serverless.yml      # Serverless configuration
├── agent.py            # LangGraph agent with browser toolkit
├── pyproject.toml      # Python dependencies
├── Dockerfile          # Container configuration
├── test-invoke.py      # Test script
└── README.md           # This file

Quick Start

1. Deploy

serverless deploy

2. Note the Runtime Endpoint URL

After deployment, note the runtime endpoint URL from the output (the ARN is embedded in the URL path):

ai:
  agents:
    browserAgent: https://bedrock-agentcore.us-east-1.amazonaws.com/runtimes/arn:aws:bedrock-agentcore:us-east-1:123456789012:runtime/xxx/invocations

3. Test

export RUNTIME_ARN="arn:aws:bedrock-agentcore:us-east-1:123456789012:runtime/xxx"
python test-invoke.py

How It Works

Browser Toolkit

The example uses langchain_aws.tools.create_browser_toolkit which provides these tools:

Tool	Description
`navigate_browser`	Navigate to a URL
`click_element`	Click on an element using CSS selector
`type_text`	Type text into an input field
`extract_text`	Extract all text content from the page
`extract_hyperlinks`	Extract all hyperlinks from the page
`get_elements`	Get elements matching a CSS selector
`current_webpage`	Get the current page URL and title
`navigate_back`	Go back to the previous page
`take_screenshot`	Take a screenshot of the page
`scroll_page`	Scroll the page in a direction
`wait_for_element`	Wait for an element to appear

LangGraph Agent

from langchain.chat_models import init_chat_model
from langchain_aws.tools import create_browser_toolkit
from langgraph.prebuilt import create_react_agent

# Create toolkit
toolkit, browser_tools = create_browser_toolkit(region="us-east-1")

# Initialize chat model (default: global.anthropic.claude-sonnet-5, override via MODEL_ID)
llm = init_chat_model(
    MODEL_ID,
    model_provider="bedrock_converse",
)

# Create agent with browser tools
agent = create_react_agent(
    model=llm,
    tools=browser_tools,
)

# Run with session isolation
config = {"configurable": {"thread_id": "session-123"}}
result = await agent.ainvoke(
    {"messages": [{"role": "user", "content": "Navigate to example.com"}]},
    config=config
)

Session Isolation

Each thread_id gets its own browser session, enabling concurrent usage:

# Each thread gets its own browser session
config_user1 = {"configurable": {"thread_id": "user-1"}}
config_user2 = {"configurable": {"thread_id": "user-2"}}

Example Prompts

Navigation:

Navigate to https://example.com and tell me the main heading

Extract Links:

Navigate to https://aws.amazon.com and extract the first 5 hyperlinks

Page Analysis:

Navigate to https://python.org and describe the main sections

Form Interaction:

Navigate to https://google.com, type "AWS Lambda" in the search box

Dependencies

langchain-aws - AWS integrations for LangChain
langgraph - Agent orchestration
playwright - Browser automation (required by browser toolkit)
beautifulsoup4 - HTML parsing

Reference

Cleanup

serverless remove

Next Steps

Gateway Example - Add custom Lambda tools
Memory Example - Add conversation persistence
Browser Documentation - Full configuration reference