ManyAgents is a native macOS application designed to manage and coordinate multiple AI coding agents within a single window. It allows users to run a team of Claude Code and Codex agents simultaneously, with each agent operating in its own tab. The application is intended for developers who want to delegate software development tasks to multiple agents while maintaining oversight and control from one interface. The workflow centers around an orchestrator agent that receives a high-level request from the user and breaks it into discrete tasks. Worker agents each handle individual tasks in their own git worktrees to avoid conflicts, then test the results and report back. Thinker agents are called in for complex problems such as design decisions or bugs that span multiple components. Reviewer agents can inspect completed work from a read only perspective and use a separate AI provider when Codex is configured. A built-in browser runs alongside the chat, allowing agents to interact with the application being built—clicking through flows, filling forms, and checking results. Agents can also display an iOS Simulator within the same window. When an agent requires a human decision, the request appears in a dedicated section with the agent’s recommendation, ensuring nothing is left unseen in background tabs. ManyAgents is free for 14 days without requiring an account or credit card. It requires macOS 14 or later and a Claude Code account. Codex integration is optional. The application does not incur additional AI charges beyond the user’s existing Claude Code and Codex plans, and all code remains local on the user’s machine. Version 1.2.7 is the current release.