aws-samples/amazon-bedrock-client-for-mac

What it solves

This project provides a native macOS application that allows users to interact with Amazon Bedrock foundation models directly from their desktop. It eliminates the need for a web-based console, offering a more integrated experience for conversations, document analysis, and image generation while keeping conversation history and settings stored locally on the user's Mac.

How it works

The app acts as a client that connects directly to the AWS Bedrock API using the user's existing AWS profiles or API keys. It supports multiple model types (text, image, video, and embeddings) and manages local storage for chats, drafts, and settings. It integrates with the Model Context Protocol (MCP) to support local stdio and HTTP servers, and includes built-in tools for interacting with the local file system, Git, and the shell.

Who it’s for

Developers and AWS users on macOS who want a high-performance, native interface for accessing Bedrock models with local control over their data and tool integration.

Highlights

• Native macOS Interface: Built with Swift, featuring a frosted sidebar, system-wide appearance modes, and a centered composer.
• Local-First Storage: Conversations, skills, and settings are stored locally on the Mac, with no hosted backend or account system.
• Local Tooling & MCP: Includes built-in tools for shell commands, file operations, and Git, plus support for MCP connections to extend functionality.
• Model Flexibility: Ability to search, favorite, and switch models within a single conversation while maintaining context.
• Advanced Conversation Management: Supports message queuing, rich formatting for responses, and local search across conversations.
• Demo Library: Provides pre-configured prompts for various tasks including reasoning, vision, and model comparison.