Report Abuse

Basic Information

Bedrock Engineer is a native desktop application and sample project that provides autonomous AI agent capabilities built on Amazon Bedrock. It is designed to assist software development and other productivity tasks by letting users create, customize, and run agents that can read and edit files, execute commands, run code, search the web, and interact with knowledge bases. The repository includes a downloadable macOS and Windows app, build instructions using npm, configuration paths, and guidance for resolving platform-specific issues. It also integrates with external tooling such as an Agent Preparation Toolkit and Model Context Protocol (MCP) clients to extend agent capabilities. The project bundles multi-agent management, tool customization, a curated agent directory, voice chat, and content generation features so practitioners can deploy and iterate on agent-driven workflows.

Links

Categorization

App Details

Features
Interactive Agent Chat that uses Amazon Bedrock models and supports multi-agent workflows and customization. File system tools for creating, reading, writing, moving, copying, and listing project files with Excel-to-CSV conversion. Web and site tools powered by a Tavily search integration and a fetchWebsite tool that handles large content chunking. Amazon Bedrock integrations for image, video, and recognition tasks plus invokeBedrockAgent, retrieve from Bedrock Knowledge Base, and invokeFlow for custom pipelines. Secure code execution via a Python codeInterpreter running in Docker, executeCommand for controlled shell execution, and screen/camera capture tools. Background Agent scheduling with cron expressions, an Agent Directory for sharing and discovering agents, Nova Sonic real-time voice chat, website/diagram/Step Functions generators, and support for MCP and inference profiling.
Use Cases
The repository helps developers and teams accelerate development by automating routine tasks such as project scaffolding, code generation, analysis, and refactoring. Agents can access local project files, run controlled commands, and execute Python analyses in isolated Docker environments to protect security while enabling data science workflows. Web search and knowledge-base retrieval let agents incorporate up-to-date information when generating code or documentation. Background scheduling and session continuity enable recurring automation and monitoring workflows. The Agent Directory and organization sharing streamline reuse and collaboration. Multimedia generation and recognition, voice chat, and diagram/website generators make the tool usable beyond code tasks for design, documentation, and prototyping. MCP and APT support extensibility with external tools and custom flows.

Please fill the required fields*