AWS Open-Sources Kiro Crew for Multi-Agent Coding
AWS has open-sourced Kiro Crew, an orchestration system that lets developers run multiple asynchronous AI coding agents concurrently to automate background tasks like debugging and migrations.

Amazon Web Services released Kiro Crew under the Apache 2.0 license, making the tool available for macOS, Linux, and Windows. Originally developed internally at Amazon under the name MeshClaw, the platform was organically adopted by more than 39,000 internal developers and 500 contributors over a six-month period. Created by Amazon engineers Bolin Chen, Zejiang Guo, and Zezhen Xu, the project aims to let developers initiate complex tasks and walk away rather than constantly monitoring single prompts.
Kiro Crew orchestrates concurrent agents using the Agent Client Protocol (ACP), allowing them to retain project context across sessions through shared memory. The system supports asynchronous operations such as ticket triage, pull request monitoring, incident investigation, and CI/CD migrations. It integrates with external tools via Model Context Protocol (MCP) and webhooks, and it runs on the Kiro CLI using existing .kiro configurations. To ensure security, the platform includes an OS-level sandbox, denied-by-default commands, credential redaction, and signed audit logs.
For software practitioners, Kiro Crew introduces a control plane that shifts AI assistance from active chat prompting to background automation. Developers can delegate parallel jobs to subagents without blocking their primary workspace, saving time on redundant prompt engineering. However, early adopters have raised concerns regarding operational costs, noting that Kiro Crew can consume API tokens significantly faster than the standard Kiro CLI. The platform also features built-in integrations for communication tools like Slack, Telegram, and WeCom to provide live updates on agent activity.
This is our own summary of reporting by InfoQ AI


