Aws Bedrock Guardrails MCP

🚀 Bedrock Guardrails MCP Server

This MCP server provides all AWS Bedrock Guardrails management features as tools. It allows you to programmatically create, edit, delete, and export guardrails (including all policy types), or use LLMs like Claude Desktop. It also supports exporting guardrail configurations as Terraform files for infrastructure - as - code workflows.

🚀 Quick Start

This Bedrock Guardrails MCP Server exposes all AWS Bedrock Guardrails management features, enabling you to manage guardrails programmatically or via LLMs. Follow the steps below to get started.

✨ Features

• Create, update, delete, and list AWS Bedrock Guardrails.
• Fully support all policy types: content, topic, word, sensitive information, contextual grounding, etc.
• Export any guardrail as a Terraform .tf file (CloudFormation schema).
• Secure authentication using AWS credentials from environment variables.
• Ready for integration with Claude Desktop or any MCP - compatible client.

📦 Installation

Prerequisites

• Python 3.12+
• uv package manager
• AWS account with Bedrock Guardrails permissions
• AWS credentials set in your environment (see below)

Setup Steps

1. Clone the repository
2. Install dependencies:

   uv venv
   source .venv/bin/activate
   uv pip install -r requirements.txt  # or use pyproject.toml with uv
   

3. Set AWS credentials in your environment:

   export AWS_ACCESS_KEY_ID=your - access - key
   export AWS_SECRET_ACCESS_KEY=your - secret - key
   export AWS_SESSION_TOKEN=your - session - token  # if using temporary credentials
   export AWS_REGION=us - east - 1  # or your preferred region
   

   ⚠️ Important Note

   Never commit your credentials to source control. Use environment variables or a secure credentials manager.

4. Run the server:

   uv run server.py
   

📚 Documentation

Claude Desktop Integration

To use this server as a tool in Claude Desktop, add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "BedrockGuardrailsMCP": {
      "command": "uv",
      "args": [
        "--directory",
        "/Users/haji/mcp - servers/bedrock - guardrails - mcp",
        "run",
        "server.py"
      ]
    }
  }
}

• Replace the directory path with your actual project location.
• Restart Claude Desktop after updating the config.

💻 Usage Examples

Basic Usage

• Create a guardrail with full policy config: Use the create_guardrail_full tool, passing all required and optional policy configs as dictionaries.
• Update a guardrail: Use update_guardrail_full with the guardrail ID and any fields to update.
• Export to Terraform: Use export_guardrail_to_terraform(guardrail_id) to get a .tf resource block for your guardrail.

🔧 Technical Details

Impact & Use Cases

• Enterprise LLM Safety: Centrally manage and automate Bedrock Guardrails for all your generative AI applications.
• DevOps & IaC: Export guardrails to Terraform for version control, reproducibility, and CI/CD integration.
• LLM Tooling: Enable LLMs (like Claude) to programmatically manage safety policies, audit, and remediate risks.

Security

• Credentials are only read from environment variables and never stored in code or logs.
• All AWS API calls are made using the official boto3 SDK.

📄 License

See LICENSE file.