AWS Authentication Guide

Complete guide for setting up AWS credentials and authentication for Amazon Bedrock access.

Table of contents

  1. Overview
  2. Authentication Methods
    1. 1. Static Credentials
    2. 2. AWS Profile
    3. 3. IAM Roles (Recommended for Production)
    4. 4. Role Assumption
    5. 5. Web Identity Tokens (OIDC)
  3. IAM Permissions
    1. Required Permissions
    2. Model-Specific Permissions
    3. Enhanced Permissions
  4. Setup Examples
    1. Development Environment
    2. Production Deployment
  5. Regional Configuration
    1. Available Regions
    2. Model Availability by Region
  6. Troubleshooting
    1. Common Issues
    2. Debugging Commands
  7. Security Best Practices
    1. 1. Principle of Least Privilege
    2. 2. Use IAM Roles Instead of Keys
    3. 3. Credential Rotation
    4. 4. Environment-Specific Configuration
    5. 5. Audit and Monitoring
  8. Configuration Examples
    1. Docker Deployment
    2. Docker Compose
    3. Kubernetes Deployment
    4. GitHub Actions

Overview

The Open Bedrock Server Server supports multiple AWS authentication methods to access Amazon Bedrock models. This guide covers all available authentication options and best practices for different deployment scenarios.

Authentication Methods

1. Static Credentials

The simplest method using direct access keys.

Configuration:

AWS_ACCESS_KEY_ID=AKIA1234567890EXAMPLE
AWS_SECRET_ACCESS_KEY=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY
AWS_REGION=us-east-1

Use Cases:

  • Development and testing
  • Simple deployments
  • When other methods aren’t available

Security Considerations:

  • Store credentials securely
  • Use least-privilege IAM policies
  • Rotate keys regularly
  • Never commit to version control

2. AWS Profile

Use AWS CLI profiles for credential management.

Setup:

# Configure AWS CLI profile
aws configure --profile my-profile
# AWS Access Key ID: AKIA1234567890EXAMPLE
# AWS Secret Access Key: wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY
# Default region name: us-east-1
# Default output format: json

Configuration:

AWS_PROFILE=my-profile
AWS_REGION=us-east-1

Use Cases:

  • Development environments
  • Multiple AWS accounts
  • Shared development machines

Use IAM roles for secure, temporary credentials.

For EC2 Instances:

# No credentials needed - uses instance profile
AWS_REGION=us-east-1

For ECS/Fargate:

# Uses task role
AWS_REGION=us-east-1

For Lambda:

# Uses execution role
AWS_REGION=us-east-1

4. Role Assumption

Assume roles for cross-account access or enhanced security.

Configuration:

AWS_ROLE_ARN=arn:aws:iam::123456789012:role/BedrockAccessRole
AWS_ROLE_SESSION_NAME=amazon-chat-session
AWS_REGION=us-east-1

# Optional: External ID for enhanced security
AWS_EXTERNAL_ID=unique-external-id

# Optional: Base credentials for assumption
AWS_ACCESS_KEY_ID=AKIA1234567890EXAMPLE
AWS_SECRET_ACCESS_KEY=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY

5. Web Identity Tokens (OIDC)

For Kubernetes, GitHub Actions, and other OIDC providers.

Configuration:

AWS_ROLE_ARN=arn:aws:iam::123456789012:role/GitHubActionsRole
AWS_WEB_IDENTITY_TOKEN_FILE=/var/run/secrets/eks.amazonaws.com/serviceaccount/token
AWS_REGION=us-east-1

IAM Permissions

Required Permissions

Minimum IAM policy for Bedrock access:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "bedrock:InvokeModel",
                "bedrock:InvokeModelWithResponseStream"
            ],
            "Resource": [
                "arn:aws:bedrock:*::foundation-model/anthropic.claude-*",
                "arn:aws:bedrock:*::foundation-model/amazon.titan-*",
                "arn:aws:bedrock:*::foundation-model/ai21.*",
                "arn:aws:bedrock:*::foundation-model/cohere.*",
                "arn:aws:bedrock:*::foundation-model/meta.*"
            ]
        }
    ]
}

Model-Specific Permissions

For specific models only:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "bedrock:InvokeModel",
                "bedrock:InvokeModelWithResponseStream"
            ],
            "Resource": [
                "arn:aws:bedrock:us-east-1::foundation-model/anthropic.claude-3-haiku-20240307-v1:0",
                "arn:aws:bedrock:us-east-1::foundation-model/anthropic.claude-3-sonnet-20240229-v1:0"
            ]
        }
    ]
}

Enhanced Permissions

For additional features like model listing:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "bedrock:InvokeModel",
                "bedrock:InvokeModelWithResponseStream",
                "bedrock:ListFoundationModels",
                "bedrock:GetFoundationModel"
            ],
            "Resource": "*"
        }
    ]
}

Setup Examples

Development Environment

Option 1: AWS CLI Profile

# Install AWS CLI
pip install awscli

# Configure profile
aws configure --profile bedrock-dev
# Enter your credentials and region

# Set environment
export AWS_PROFILE=bedrock-dev
export AWS_REGION=us-east-1

# Start server
bedrock-chat serve

Option 2: Environment Variables

# Set credentials
export AWS_ACCESS_KEY_ID=your-access-key
export AWS_SECRET_ACCESS_KEY=your-secret-key
export AWS_REGION=us-east-1

# Start server
bedrock-chat serve

Production Deployment

EC2 Instance with IAM Role:

  1. Create IAM role with Bedrock permissions
  2. Attach role to EC2 instance
  3. Configure application:
# .env file
AWS_REGION=us-east-1
# No credentials needed - uses instance profile

ECS/Fargate with Task Role:

# ECS task definition
taskRoleArn: arn:aws:iam::123456789012:role/BedrockTaskRole
containerDefinitions:
  - name: amazon-chat-server
    environment:
      - name: AWS_REGION
        value: us-east-1

Kubernetes with IRSA:

# ServiceAccount
apiVersion: v1
kind: ServiceAccount
metadata:
  name: amazon-chat-sa
  annotations:
    eks.amazonaws.com/role-arn: arn:aws:iam::123456789012:role/BedrockRole

---
# Deployment
apiVersion: apps/v1
kind: Deployment
metadata:
  name: amazon-chat-server
spec:
  template:
    spec:
      serviceAccountName: amazon-chat-sa
      containers:
      - name: server
        env:
        - name: AWS_REGION
          value: us-east-1

Regional Configuration

Available Regions

Bedrock is available in specific regions:

# US East (N. Virginia)
AWS_REGION=us-east-1

# US West (Oregon)
AWS_REGION=us-west-2

# Europe (Frankfurt)
AWS_REGION=eu-central-1

# Asia Pacific (Singapore)
AWS_REGION=ap-southeast-1

# Asia Pacific (Tokyo)
AWS_REGION=ap-northeast-1

Model Availability by Region

Different models are available in different regions:

Model us-east-1 us-west-2 eu-central-1 ap-southeast-1
Claude 3 Haiku ✅ ✅ ✅ ✅
Claude 3 Sonnet ✅ ✅ ✅ ✅
Claude 3 Opus ✅ ✅ ❌ ❌
Titan Text Express ✅ ✅ ✅ ✅

Troubleshooting

Common Issues

1. Access Denied Errors

Error: An error occurred (AccessDeniedException) when calling the InvokeModel operation

Solutions:

  • Verify IAM permissions include bedrock:InvokeModel
  • Check resource ARNs in policy
  • Ensure model is available in your region

2. Invalid Credentials

Error: The security token included in the request is invalid

Solutions:

  • Check AWS credentials are correct
  • Verify credentials haven’t expired
  • For assumed roles, check role trust policy

3. Region Mismatch

Error: Model not found in region

Solutions:

  • Verify model availability in your region
  • Check AWS_REGION environment variable
  • Use correct model IDs for your region

4. Role Assumption Failures

Error: User is not authorized to perform: sts:AssumeRole

Solutions:

  • Check role trust policy allows your principal
  • Verify external ID if required
  • Ensure base credentials have sts:AssumeRole permission

Debugging Commands

Test AWS Credentials:

aws sts get-caller-identity

List Available Models:

aws bedrock list-foundation-models --region us-east-1

Test Model Access:

aws bedrock invoke-model \
  --model-id anthropic.claude-3-haiku-20240307-v1:0 \
  --body '{"anthropic_version":"bedrock-2023-05-31","max_tokens":10,"messages":[{"role":"user","content":"Hi"}]}' \
  --cli-binary-format raw-in-base64-out \
  --region us-east-1 \
  output.json

Security Best Practices

1. Principle of Least Privilege

Grant only necessary permissions:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "bedrock:InvokeModel"
            ],
            "Resource": [
                "arn:aws:bedrock:us-east-1::foundation-model/anthropic.claude-3-haiku-20240307-v1:0"
            ],
            "Condition": {
                "StringEquals": {
                    "aws:RequestedRegion": "us-east-1"
                }
            }
        }
    ]
}

2. Use IAM Roles Instead of Keys

Prefer IAM roles over static credentials:

# Good: Use instance profile
# No credentials in environment

# Avoid: Static credentials
export AWS_ACCESS_KEY_ID=...
export AWS_SECRET_ACCESS_KEY=...

3. Credential Rotation

For static credentials, implement rotation:

import boto3
from datetime import datetime, timedelta

def check_key_age():
    iam = boto3.client('iam')
    
    # Get access key metadata
    response = iam.list_access_keys()
    
    for key in response['AccessKeyMetadata']:
        age = datetime.now(key['CreateDate'].tzinfo) - key['CreateDate']
        
        if age > timedelta(days=90):
            print(f"Warning: Access key {key['AccessKeyId']} is {age.days} days old")

4. Environment-Specific Configuration

Use different credentials for different environments:

# Development
export AWS_PROFILE=dev-profile

# Staging
export AWS_PROFILE=staging-profile

# Production
# Use IAM roles, no profile needed

5. Audit and Monitoring

Monitor AWS API usage:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "bedrock:InvokeModel"
            ],
            "Resource": "*",
            "Condition": {
                "Bool": {
                    "aws:CloudTrailLogged": "true"
                }
            }
        }
    ]
}

Configuration Examples

Docker Deployment 

FROM python:3.11-slim

# Install application
COPY . /app
WORKDIR /app
RUN pip install -e .

# Use IAM role (no credentials in image)
ENV AWS_REGION=us-east-1

CMD ["amazon-chat", "serve", "--host", "0.0.0.0"]

Docker Compose 

version: '3.8'
services:
  amazon-chat:
    build: .
    environment:
      - AWS_REGION=us-east-1
      # For development only
      - AWS_ACCESS_KEY_ID=${AWS_ACCESS_KEY_ID}
      - AWS_SECRET_ACCESS_KEY=${AWS_SECRET_ACCESS_KEY}
    ports:
      - "8000:8000"

Kubernetes Deployment 

apiVersion: apps/v1
kind: Deployment
metadata:
  name: amazon-chat-server
spec:
  replicas: 3
  selector:
    matchLabels:
      app: amazon-chat-server
  template:
    metadata:
      labels:
        app: amazon-chat-server
    spec:
      serviceAccountName: amazon-chat-sa
      containers:
      - name: server
        image: amazon-chat-server:latest
        env:
        - name: AWS_REGION
          value: us-east-1
        ports:
        - containerPort: 8000

GitHub Actions 

name: Deploy
on:
  push:
    branches: [main]

jobs:
  deploy:
    runs-on: ubuntu-latest
    permissions:
      id-token: write
      contents: read
    
    steps:
    - uses: actions/checkout@v3
    
    - name: Configure AWS credentials
      uses: aws-actions/configure-aws-credentials@v2
      with:
        role-to-assume: arn:aws:iam::123456789012:role/GitHubActionsRole
        aws-region: us-east-1
    
    - name: Test Bedrock access
      run: |
        aws bedrock list-foundation-models --region us-east-1

This guide covers all aspects of AWS authentication for the Open Bedrock Server Server. Choose the method that best fits your deployment scenario and security requirements.

Open Bedrock Server is an open-source project licensed under MIT. Not affiliated with or endorsed by OpenAI or AWS.

This site uses Just the Docs, a documentation theme for Jekyll.