Manage parallel coding agents with Agent of Empires

Use this skill when an operator needs one control plane for several coding agents working across branches, worktrees, or repositories. The agent should create isolated sessions, attach the right coding CLI, monitor whether each session is running or waiting for input, surface diffs for review, and keep long-running work recoverable through tmux-backed sessions. The workflow is strongest when the operator is coordinating Claude Code, OpenCode, Codex CLI, Gemini CLI, Cursor CLI, Copilot CLI, or similar tools and needs a status dashboard rather than many disconnected terminals.\n\nInvoke this instead of using each agent product normally when the job involves parallel execution, mobile/web check-ins, sandboxed experiments, or persistent sessions that must survive terminal disconnects. The scope boundary is session orchestration, visibility, and review for local coding agents. The skill should not describe Agent of Empires as a replacement model or generic developer platform; it should produce a concrete operating flow: install the CLI, create or attach sessions, optionally enable Docker sandboxing, monitor state, inspect diffs, and hand control back to the human for approval.