Sample Agentic AI Web

v1.0.0Coding Agentsstable

This project demonstrates how to use AWS Bedrock with Anthropic Claude and Amazon Nova models to create a web automation assistant with tool use, human-in-the-loop interaction, and vision capabilities.

agentic-aiaiawscheckbotgenai
Share:
50
Stars
0
Downloads
0
Weekly
0/5

What is Sample Agentic AI Web?

Sample Agentic AI Web is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to this project demonstrates how to use aws bedrock with anthropic claude and amazon nova models to create a web automation assistant with tool use, human-in-the-loop interaction, and vision capabilities...

This project demonstrates how to use AWS Bedrock with Anthropic Claude and Amazon Nova models to create a web automation assistant with tool use, human-in-the-loop interaction, and vision capabilities.

This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • This project demonstrates how to use AWS Bedrock with Anthro

Use Cases

AWS Bedrock web automation
Vision-based browser control with human-in-the-loop
aws-samples

Maintainer

LicenseMIT-0
Languagepython
Versionv1.0.0
UpdatedApr 9, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx sample-agentic-ai-web

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Sample Agentic AI Web

Sample Agentic AI Web is an AWS sample project that demonstrates how to build a web automation assistant using AWS Bedrock with Anthropic Claude and Amazon Nova models, Playwright for browser control, and the Model Context Protocol for tool integration. It progressively adds capabilities across 11 example steps—from basic API calls to a full MCP client/server architecture—including screenshot capture, visual element interaction, form filling, and human-in-the-loop checkpoints. It serves as a practical blueprint for developers building agentic AI systems that need to browse, analyze, and interact with web interfaces on behalf of users.

Prerequisites

  • Python 3.11 or later installed
  • AWS CLI configured with credentials that have access to Amazon Bedrock (run `aws configure`)
  • Amazon Bedrock enabled in your AWS account with access to Claude 3 models and Amazon Nova models granted
  • Playwright installed: `pip install playwright && playwright install`
  • An MCP-compatible client or the ability to run the included MCP client script directly
1

Clone the repository

Clone the AWS sample repository to your local machine.

git clone https://github.com/aws-samples/sample-agentic-ai-web.git
cd sample-agentic-ai-web
2

Create a virtual environment and install dependencies

Use a Python virtual environment to isolate the project's dependencies.

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
pip install -r requirements.txt
3

Install Playwright browsers

Playwright needs to download browser binaries on first use.

playwright install chromium
4

Configure AWS credentials for Bedrock access

Ensure your AWS credentials are configured and that you have enabled model access in the Bedrock console for Claude 3 and Amazon Nova models in your target region.

aws configure
# Set your Access Key ID, Secret Access Key, region (e.g. us-east-1), and output format
5

Run the progression of examples

The project is structured as numbered steps (01 through 11). Start with the simplest example and work up to the full MCP-based agentic web assistant.

# Start with step 01 (basic Bedrock API call):
python 01-no-tools/solution.py

# Run step 11 (full MCP client with web automation):
python 11-mcp-client/solution.py
6

Register the MCP server with your client (step 10-11)

For the MCP server steps, add the web automation server to your MCP client configuration pointing to the Python entry point.

{
  "mcpServers": {
    "web-automation": {
      "command": "python",
      "args": ["/path/to/sample-agentic-ai-web/10-mcp-server/server.py"]
    }
  }
}

Sample Agentic AI Web Examples

Client configuration

MCP client configuration for the web automation server in the sample project.

{
  "mcpServers": {
    "agentic-web": {
      "command": "python",
      "args": ["/path/to/sample-agentic-ai-web/10-mcp-server/server.py"],
      "env": {
        "AWS_DEFAULT_REGION": "us-east-1"
      }
    }
  }
}

Prompts to try

Once the web automation server is running, you can issue browser control commands through your MCP client.

- "Navigate to https://news.ycombinator.com, take a screenshot, and summarize the top 5 stories"
- "Go to https://example.com and fill in the contact form with my name and email"
- "Search for 'AWS Bedrock MCP' on Google and scroll down to see the results"
- "Take a screenshot of the current page and describe the UI layout"

Troubleshooting Sample Agentic AI Web

Bedrock API call fails with 'AccessDeniedException'

You must explicitly enable model access in the AWS Bedrock console under 'Model access'. Navigate to the Bedrock console in your target region, click 'Model access', and request access to the Claude and Nova models you intend to use.

Playwright raises 'BrowserType.launch: Executable doesn't exist' error

Run `playwright install chromium` (or `playwright install` for all browsers) in the activated virtual environment to download the browser binaries.

Screenshots are not saved or are saved to an unexpected location

The sample saves screenshots with random UUID filenames in the working directory where the script is run. Ensure you have write permissions in that directory, or update the file path in the script to an absolute path of your choosing.

Frequently Asked Questions about Sample Agentic AI Web

What is Sample Agentic AI Web?

Sample Agentic AI Web is a Model Context Protocol (MCP) server that this project demonstrates how to use aws bedrock with anthropic claude and amazon nova models to create a web automation assistant with tool use, human-in-the-loop interaction, and vision capabilities. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Sample Agentic AI Web?

Follow the installation instructions on the Sample Agentic AI Web GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with Sample Agentic AI Web?

Sample Agentic AI Web works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Sample Agentic AI Web free to use?

Yes, Sample Agentic AI Web is open source and available under the MIT-0 license. You can use it freely in both personal and commercial projects.

Browse More Coding Agents MCP Servers

Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "sample-agentic-ai-web": { "command": "npx", "args": ["-y", "sample-agentic-ai-web"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Sample Agentic AI Web?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides