Agent Maestro is a local orchestration tool for AI coding agents. It manages the execution of coding tasks by running them one at a time through the command line interfaces of supported providers. The tool is designed to coordinate implementation, validation, independent review, automatic fix attempts, and Git workflows within a single environment. It operates entirely on the user's machine and relies on the user's installed agent CLIs, with no silent fallback to alternative providers. The application provides a React based interface that displays validation results, review findings, fix attempts, and Git outcomes in a unified view. Users can configure a preset that supplies guidance and visible review defaults. Validation, Git mode, provider selection, and limits are maintained as separate configurable choices. Review cycles retain verdicts and finding IDs, and a reported fix is only marked as resolved after a subsequent review accepts it. The Agents page distinguishes between configured models and a completed availability check; opening the page does not automatically run a diagnostic. The typical workflow involves configuring checks and an optional independent review before execution. The tool then runs the task, inspects the result, and performs bounded fix attempts if needed. A review finding can be sent back to the original implementation session, followed by validation and review again. Providers, gates, limits, and Git mode are all configurable, though approval does not guarantee correctness. Supported implementation providers include OpenAI Codex, Claude Code, GitHub Copilot, and Gemini CLI, each requiring a compatible installed CLI, authentication, account and model access, and applicable organization policy. The tool also supports an independent review role for most providers, with the exception of Antigravity, which cannot enforce the required read only reviewer boundary.