Guide · Agent skill
How to use the AWS AgentCore Cloud Browser agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The agentcore skill teaches an AI agent to run its browser automation on a cloud browser hosted by AWS Bedrock AgentCore instead of a local Chrome. The agent keeps using the same agent-browser commands, and only the location of the browser changes. This guide explains how to install the skill, trigger it with plain requests, and what it needs before it works.
Quick start
- 1Download and unzip. Download the ZIP from this page and unzip it so that SKILL.md sits in a skills folder.
- 2Place the skill folder. Copy it to ~/.claude/skills/ for personal use, or to a project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Confirm AWS credentials. Make sure AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY are set, or that your AWS profile works.
- 4Choose a region. Set AGENTCORE_REGION if you need something other than the default us-east-1.
- 5Ask for an AWS-hosted browser. Make a request such as 'use agentcore to open this page'. The agent runs the agent-browser commands on the AgentCore browser.
- 6Review the session. Open the Live View URL printed to stderr to watch the session in the AWS Console.
What the agentcore skill makes the agent do
Agent skills are folders of instructions that an AI agent loads when a task matches them. This skill covers one case: browser automation that must run on AWS infrastructure. The agent uses the standard agent-browser commands such as open, snapshot, click and screenshot. Only the place the browser runs is different.
The skill resolves AWS credentials automatically, and it can keep login state between sessions using a persistent profile.
Install the agentcore skill
Download the ZIP from this page and unzip it. The package contains a single file, SKILL.md. Put the unzipped folder in one of these locations:
- Personal install: ~/.claude/skills/
- Project install: a project's .claude/skills/ folder
- Claude apps: upload the skill under Settings → Capabilities → Skills
Check the requirements
The skill needs the agent-browser CLI, which the agent can run through Bash or npx. It also needs AWS credentials and access to AWS Bedrock AgentCore in the region you choose. The AWS CLI is optional and is only needed for SSO, named profiles or IAM role credentials.
- agent-browser CLI (via Bash or npx)
- AWS credentials: AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY, or a configured AWS profile
- AWS Bedrock AgentCore access in your chosen region
- AWS CLI (optional, for SSO, profiles or IAM roles)
Trigger the skill with natural requests
You do not need special syntax. Ask for the AWS-hosted browser in plain words. The skill is designed to match requests such as the ones below, or any task that needs AWS-hosted browser automation.
- "Use agentcore to open the pricing page and take a screenshot."
- "Run this on AWS with a cloud browser."
- "Start an agentcore session and fill in the login form."
Set credentials and region
Credentials are resolved in this order: environment variables first, then the AWS CLI fallback, which supports SSO, IAM roles and named profiles. If you already have working AWS credentials, no extra setup is needed. The default region is us-east-1, and you can change it with AGENTCORE_REGION.
- Environment variables: AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, optionally AWS_SESSION_TOKEN
- AWS CLI fallback: aws configure export-credentials
- AGENTCORE_REGION sets the AWS region (default us-east-1)
- AWS_PROFILE selects the CLI profile (default default)
aws sso login --profile my-profile
AWS_PROFILE=my-profile agent-browser -p agentcore open https://example.comKeep logins with a persistent profile
Set AGENTCORE_PROFILE_ID to store browser state such as cookies and localStorage across sessions. Log in once, then later runs start already authenticated. Only use this for sessions you are comfortable storing in AWS.
- AGENTCORE_SESSION_TIMEOUT sets session length in seconds (default 3600)
- Live View: AgentCore prints a Live View URL to stderr so you can watch the session in the AWS Console
AGENTCORE_PROFILE_ID=my-app agent-browser -p agentcore open https://app.example.com/dashboardWhat people use it for
- Running browser automation from a CI/CD pipeline or script that uses explicit AWS credentials
- Checking a web page and taking a screenshot on an AWS-hosted browser in a chosen region
- Staying logged in to an internal web app across several agent runs with a persistent profile
- Watching a long-running browser task in real time through the AWS Console Live View
Frequently asked questions
What is the agentcore skill?
It is an agent skill that tells an AI agent to run agent-browser on cloud browser sessions hosted by AWS Bedrock AgentCore. The commands stay the same as local Chrome; only the browser location changes.
Do I need an AWS account to use it?
Yes. The skill cannot authenticate without working AWS credentials, either as environment variables or through a configured AWS profile, and you need AWS Bedrock AgentCore access in your chosen region.
How do I install the agentcore skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.
Is the agentcore skill free?
The source for this skill is licensed under Apache-2.0. Using AgentCore is a separate AWS service, so check AWS pricing for your account, since a cloud browser can add cost compared with local Chrome.
Why does it say 'Failed to run aws CLI'?
That error means the AWS CLI is not installed or not in your PATH. Install it, or set AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY directly so the CLI is not needed.
What should I do when SSO credentials expire?
If you see 'AWS CLI failed: ... Run aws sso login', run aws sso login to refresh the credentials, then run the command again.
When should I use a local browser instead?
Use a local browser for simple browsing where a cloud browser adds cost and an AWS dependency for no benefit. Avoid persistent profiles for highly sensitive logins on shared infrastructure, because cookies and sessions persist in AWS.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits developers and teams whose browser automation must run on AWS, especially those already using AWS credentials, SSO or IAM roles, or who need a long-running logged-in session. If you do not have an AWS account or working credentials, or you only need simple local browsing, pick a local browser setup instead.
Download the skill