This project provides a native macOS client for Amazon Bedrock, offering a streamlined desktop experience for interacting with various AI models. It targets macOS users who need efficient access to generative AI capabilities, including text, image, and embedding models, with features like real-time streaming and multi-modal support.

How It Works

The client leverages the Unified ConverseStream API for all text generation models, ensuring consistent interaction patterns. It supports advanced features like Model Context Protocol (MCP) integration for tool use tracking and execution, and provides an enhanced reasoning capability with an increased token budget. The native macOS interface is designed for user experience, incorporating dark mode, real-time streaming responses, and robust conversation management.

Quick Start & Requirements

• Installation: Install via Homebrew: brew tap didhd/tap, brew update, brew install amazon-bedrock-client --no-quarantine. Alternatively, download the DMG.
• Prerequisites: macOS 14+, an AWS Account with Bedrock access, and configured AWS credentials (~/.aws/credentials or ~/.aws/config).
• Setup: Manual approval required in macOS System Preferences > Privacy & Security on first launch.
• Docs: Troubleshooting Guide

Highlighted Details

• Supports a wide range of Bedrock models including Claude 3.7, Llama, Mistral, DeepSeek R1, and Stable Diffusion.
• Features multi-modal capabilities with image upload for Claude 3 models.
• Includes document attachment support (PDF, Word) with rendering and image compression utilities.
• Offers AWS SSO and credential profile support for flexible authentication.

Maintenance & Community

Developed by the AWS Community. Contributions are welcomed via pull requests.

Licensing & Compatibility

Licensed under the MIT License. Permissive for commercial use and integration with closed-source applications.

Limitations & Caveats

Initial launch requires manual macOS security approval. Troubleshooting guide addresses common issues like AWS credential recognition and model access permissions.