Show HN: AgentCrew – a Markdown-first operating system for AI coding agents (github.com)

AgentCrew has been introduced as a Markdown-first operating system designed to enhance the methodology of coding agents like Claude Code and Codex by instilling a disciplined development process. Unlike conventional use cases where these AI agents operate with minimal oversight, AgentCrew promotes structured workflows that involve defining roles (such as Developer and Tester), routing tasks, and ensuring human approval before finalizing code changes. This approach addresses common pitfalls associated with AI-generated code, such as tunnel vision, skipping tests, and messy refactoring, ultimately aiming to make coding agents more reliable and collaborative in software development. Significantly, AgentCrew is not just another tool but a comprehensive system that redefines how coding agents interact with complex tasks. By implementing a layered approach to task management—differentiating between simple updates and high-risk changes—it fosters better quality control and promotes accountability. The integration is straightforward, requiring minimal setup, and effectively enhances the coding agent's capabilities without sacrificing human oversight. By ensuring that agents follow a structured process, including verification and documentation, AgentCrew intends to make the development cycle smoother, safer, and ultimately more productive.