cli-chatbot 0.1.1

A Python project for generating and using S3 presigned URLs with a chat interface.

ai-chat

A Python project for generating and using S3 presigned URLs with a chat interface powered by AWS Bedrock and Claude. This project includes:

• S3 presigned URL generation for secure upload/download operations
• File upload/download using presigned URLs
• MCP (Model Context Protocol) server for tool integration
• Chat client that interacts with Claude via LiteLLM and AWS Bedrock
• AWS integration for both S3 operations and AI model access

Features

• Generate presigned URLs for S3 bucket operations
• Upload and download files securely using presigned URLs
• Interactive chat interface with AI assistance
• Tool calling capabilities through MCP protocol
• Support for both direct Anthropic API and AWS Bedrock

Requirements

• Python 3.12+
• AWS account with appropriate permissions for:
  • S3 bucket access
  • AWS Bedrock model access (for Claude models)
• AWS credentials configured

Installation

Using pip with requirements.txt:

pip install -r requirements.txt

Using pyproject.toml:

pip install .

Configuration

Environment Variables

Create a .env file in the project root with the following variables:

# AWS Credentials for S3 and Bedrock
AWS_ACCESS_KEY_ID=your-aws-access-key-id
AWS_SECRET_ACCESS_KEY=your-aws-secret-access-key
AWS_SESSION_TOKEN=your-aws-session-token  # Optional for temporary credentials
AWS_DEFAULT_REGION=us-east-1

# LiteLLM Configuration for AWS Bedrock
LITELLM_LOG=INFO
AWS_REGION_NAME=us-east-1

# Optional: Anthropic API Key (if using direct API instead of Bedrock)
ANTHROPIC_API_KEY=your-anthropic-api-key

AWS Permissions

Ensure your AWS credentials have the following permissions:

For S3 operations:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "s3:GetObject",
                "s3:PutObject",
                "s3:GeneratePresignedUrl"
            ],
            "Resource": "arn:aws:s3:::your-bucket-name/*"
        }
    ]
}

For AWS Bedrock:

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

Usage

1. Start the MCP Server

The presigned_url.py script can function as both a library and an MCP server:

python presigned_url.py --mcp-server

2. Run the Chat Client

Start the interactive chat client that connects to the MCP server:

python client.py presigned_url.py --mcp-server

3. Using the Tools

Once the chat client is running, you can use natural language to:

• Generate presigned URLs: "Create a presigned URL for uploading to bucket 'my-bucket' with key 'file.txt'"
• Upload files: "Upload 'local-file.txt' to the presigned URL"
• Download files: "Download from the presigned URL and save as 'downloaded-file.txt'"

Model Configuration

Using AWS Bedrock (Recommended)

To use Claude models through AWS Bedrock, update the model name in client.py:

# Replace this line in client.py
model="claude-3-5-sonnet-20241022"

# With this for Bedrock
model="bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0"

Using Direct Anthropic API

If you prefer to use the direct Anthropic API, ensure you have set the ANTHROPIC_API_KEY in your .env file and use the standard model names.

Project Structure

ai-chat/
├── client.py                 # Interactive chat client
├── presigned_url.py          # Merged S3 utilities and MCP server
├── app.py                    # Additional application logic
├── requirements.txt          # Python dependencies
├── pyproject.toml           # Project configuration
├── .env                     # Environment variables (create this)
└── README.md                # This file

Troubleshooting

Common Issues

1. AWS Credentials Error: Ensure your AWS credentials are correctly set in the .env file and have the necessary permissions.

2. Bedrock Access Denied: Make sure you have enabled the Claude models in your AWS Bedrock console and have the appropriate IAM permissions.

3. S3 Bucket Access: Verify that your AWS credentials have access to the S3 bucket you're trying to use.

4. Region Mismatch: Ensure the AWS region in your .env file matches where your resources are located.

License

MIT