◈ AWS AGENTCORE BLUEPRINTS
PART III · TOOLS AND OPERATIONS · CHAPTER 17 / 36
TOCCONSULTING.FR
AMAZON BEDROCK / SERVICE DETAIL S07 (TOOLS)

AGENTCORE BROWSER

CHAPTER 17 / 36
IN ONE LINE a real web browser the agent can drive
REV 2026.07

A lot of work lives behind a website with no API: a supplier portal, a form, a dashboard. Browser gives the agent a real, managed web browser in the cloud that it can drive, click, type, read, screenshot, so it can use those sites like a person would, safely and on its own.

New to these words? Read chapter 02 "Key terms" (session, microVM, tool).
[ 1 ] WHAT IT IS

Browser is a managed Chrome browser in the cloud that the agent controls. It opens pages, clicks, fills forms, takes screenshots, and reads what is on screen, so an agent can use websites that have no API. It is built in to AgentCore and runs isolated, away from your machines.

THE AGENT GETS
A real browser it can drive step by step, plus screenshots so it can see the page.
A HUMAN CAN
Watch live and take over at any moment, then hand control back.
[ 2 ] HOW A SESSION WORKS
STEP 01

CREATE

Use the built-in browser (aws.browser.v1) or a custom one (your network, recording, permissions).

==>
STEP 02

START SESSION

Open a session: 15 minutes by default, up to 8 hours.

==>
STEP 03

DRIVE

The agent acts through the Automation stream: navigate, click, type, screenshot. Use Strands, Nova Act, or Playwright.

==>
STEP 04

WATCH

A person can open the Live View stream to watch in real time and take over.

==>
STEP 05

RECORD AND STOP

The session can be recorded to S3 for replay. Stop it to free resources.

[ 3 ] TWO LIVE STREAMS
AUTOMATION the agent acts
The agent sends actions: go to a page, click, type, screenshot.
This is how the agent does the actual work.
Driven by Strands, Nova Act, Playwright, or browser-use.
LIVE VIEW a human watches
A real-time video of the session.
A person can take control and give it back.
Good for oversight, and for steps that need a human (a login, a CAPTCHA).
[ 4 ] RECORDING, ISOLATION, AND LIMITS
RECORDING AND REPLAY
Custom browsers can be recorded to your S3 bucket (DOM changes, actions, network, console logs).
Replay them in the console, like a screen recording, for debugging and audit.
ISOLATION AND LIMITS
Each session runs in its own microVM, wiped at the end.
Default 15 minutes, up to 8 hours; up to 500 sessions at once; data kept 30 days.
Plain definitions: chapter 02 "Key terms". Hands-on code: chapter 18 "Browser, practical".
SRC: AWS AgentCore Developer Guide (browser quickstart, fundamentals, session management). DRAWN 2026.06, REV 2026.07.26
◄ PREVIOUS · CH 16
AgentCore Code Interpreter, Practical
NEXT · CH 18 ►
AgentCore Browser, Practical
AWS AGENTCORE BLUEPRINTS · CONTENTS
Back to tocconsulting.fr Cover & Table of Contents
PART I · FOUNDATIONS
01AgentCore Overview, Plain LanguageCONCEPT02AgentCore, Key Terms in Plain LanguageCONCEPT03AgentCore, Common Mistakes ExplainedCONCEPT04Inside an AgentCore Agent, BlueprintCONCEPT05Harness, Practical 1, Strands PathPRACTICAL06Harness, Practical 2, Managed HarnessPRACTICAL
PART II · CORE SERVICES
07AgentCore Runtime, Plain LanguageCONCEPT08AgentCore Runtime, PracticalPRACTICAL09AgentCore Memory, Plain LanguageCONCEPT10AgentCore Memory, PracticalPRACTICAL11AgentCore Gateway, Plain LanguageCONCEPT12AgentCore Gateway, PracticalPRACTICAL13AgentCore Identity, Plain LanguageCONCEPT14AgentCore Identity, PracticalPRACTICAL
PART III · TOOLS AND OPERATIONS
15AgentCore Code Interpreter, Plain LanguageCONCEPT16AgentCore Code Interpreter, PracticalPRACTICAL17AgentCore Browser, Plain LanguageCONCEPT18AgentCore Browser, PracticalPRACTICAL19AgentCore Observability, Plain LanguageCONCEPT20AgentCore Observability, PracticalPRACTICAL21AgentCore Evaluations, Plain LanguageCONCEPT22AgentCore Evaluations, PracticalPRACTICAL
PART IV · GOVERNANCE AND TRANSACTIONS
23AgentCore Policy, Plain LanguageCONCEPT24AgentCore Policy, PracticalPRACTICAL25AgentCore Registry, Plain LanguageCONCEPT26AgentCore Registry, PracticalPRACTICAL27AgentCore Payments, Plain LanguageCONCEPT28AgentCore Payments, PracticalPRACTICAL
PART V · MULTI-AGENT AND ARCHITECTURE
29A2A and Multi-Agent, Plain LanguageCONCEPT30A2A and Multi-Agent, PracticalPRACTICAL31Architecture Patterns, Plain LanguageCONCEPT32Multi-Tenant Agents, Plain LanguageCONCEPT33Agent Security and Threat Model, Plain LanguageCONCEPT34Governance at Org Scale, Plain LanguageCONCEPT35Governance at Org Scale, PracticalPRACTICAL36Multi-Region and Distribution, Plain LanguageCONCEPT