Agent of Empires – Multi-Agent Coding Session Manager
Run Claude Code, OpenCode, and a dozen other AI coding agents in parallel from a single TUI or web dashboard. Each agent gets its own tmux session, git worktree, and optional Docker sandbox.
TL;DR
TL;DR: Agent of Empires is a self-hosted session manager that runs multiple AI coding agents (Claude Code, OpenCode, Codex CLI, and more) in parallel from a single terminal or browser dashboard, keeping each agent in its own persistent tmux session with optional Docker sandboxing.
Source and Accuracy Notes
⚠️ This section is MANDATORY. All links must be verified from actual source, not guessed.
- Project page: agent-of-empires.com
- Source repository: github.com/njbrake/agent-of-empires
- License: MIT (verified via
LICENSEfile in repository, Copyright (c) 2026 Nathan Brake) - HN launch thread: news.ycombinator.com/item?id=46588905
- Docs: agent-of-empires.com/docs/features/
What Is Agent of Empires?
Agent of Empires (AoE) is an open-source session manager for AI coding agents on Linux and macOS. It solves the chaos of running multiple AI agents simultaneously: each agent runs in its own isolated tmux session, so closing your terminal, losing your SSH connection, or crashing your GUI does not kill the agent mid-task.
From the README:
Run many agents in parallel across different branches, each in its own isolated session with optional Docker sandboxing. Keeping track of which agent is stuck, which is waiting on input, and which just made a mess of your working tree becomes a part-time job; AoE makes it a glance: one dashboard, one status column, git worktrees and sandboxes set up for you, and sessions that outlive your terminal, reachable from your laptop, phone, or tablet.
Supported Agents
AoE supports a wide range of AI coding agents out of the box:
- Claude Code
- OpenCode
- Mistral Vibe
- Codex CLI
- Gemini CLI
- Antigravity CLI
- Cursor CLI
- Copilot CLI
- Pi.dev
- Factory Droid
- Hermes
- Kiro CLI
- Qwen Code
- Kimi Code
Setup Workflow
Prerequisites
Step 1: Install AoE
macOS (Homebrew):
brew install aoe
Linux / manual install:
curl -fsSL https://raw.githubusercontent.com/agent-of-empires/agent-of-empires/main/install.sh | bash
From source:
git clone https://github.com/njbrake/agent-of-empires.git
cd agent-of-empires
cargo build --release
./target/release/aoe
Step 2: Start the TUI Dashboard
aoe
This opens the terminal dashboard showing all active agent sessions and their current status (running, waiting for input, idle).
Step 3: Launch an Agent
From the TUI, select a repo and agent type. AoE will:
- Create a git worktree for the branch (or use the main branch)
- Spin up a new tmux session for the agent
- Optionally launch the agent inside a Docker container
Step 4: Access the Web Dashboard
From the TUI, press R to expose the web dashboard over HTTPS. This uses Tailscale Funnel or Cloudflare Tunnel to give you remote access with QR code + passphrase authentication.
The web dashboard renders each agent’s terminal in the browser, lets you switch sessions, type into the terminal, and review diffs — all from a mobile browser.
Deeper Analysis
Session Persistence
Each agent runs inside tmux, which means sessions persist beyond the lifetime of your terminal emulator or SSH connection. This is critical for long-running refactors or tests that would otherwise time out.
The key tmux shortcut: Ctrl+b d detaches from a session and returns to the AoE TUI without killing the agent.
Docker Sandboxing
With Docker enabled, each agent runs in an isolated container with shared authentication volumes. This prevents a misbehaving agent from corrupting your host filesystem. Podman and Apple Containers are also supported.
Multi-Repo Workspaces
AoE can drive agents across multiple git repositories simultaneously, each in its own worktree. This is useful for testing coordinated changes across a monorepo or a set of related services.
CLI and HTTP API
For programmatic control, AoE exposes both a CLI and an HTTP API. External orchestrators can start, stop, and query agent sessions without touching the TUI.
Practical Evaluation Checklist
- [ ] tmux is installed and working (
tmux -V) - [ ] AoE installs via Homebrew or curl script
- [ ] TUI dashboard launches and shows agent list
- [ ] Can start an agent in a git worktree
- [ ] Agent session persists after terminal close
- [ ] Web dashboard accessible from mobile browser
- [ ] Docker sandbox spins up correctly (if Docker is installed)
- [ ] Session resume works after AoE restart
Security Notes
- The web dashboard exposes a local server over HTTPS. Use the passphrase and QR code authentication when enabling remote access.
- Docker sandboxing is strongly recommended if running untrusted agent code.
- API keys and credentials should be stored outside the agent worktree and mounted as read-only volumes where possible.
- AoE stores session data under
~/.aoe/— ensure filesystem permissions restrict access to your user account.
FAQ
Q: Does AoE work on Windows? A: AoE targets Linux and macOS. Windows Subsystem for Linux (WSL2) may work but is not officially supported.
Q: How does this differ from running agents in plain tmux? A: AoE automates worktree creation, Docker sandbox setup, and provides a structured status dashboard. Without AoE, you manage each tmux session manually and have no unified view of which agent is waiting, running, or stuck.
Q: Can I use AoE with agents I run myself (not on the supported list)? A: Yes. AoE has profiles, repo config, and agent overrides that let you point to custom agent launchers. The supported list covers the major ones out of the box.
Q: Does AoE support LlamaIndex or LangChain agents? A: AoE is a session and orchestration layer for CLI-based coding agents. It does not directly integrate LlamaIndex or LangChain — those are used inside the agents it manages, not as separate runtime targets.
Conclusion
Agent of Empires fills a real gap for developers who work with multiple AI coding agents simultaneously. Instead of juggling terminal tabs and losing track of which agent is mid-task, you get one structured dashboard with persistent sessions, optional sandboxing, and remote access from any browser.
If you run Claude Code, OpenCode, or any of the supported agents for serious work, AoE is worth adding to your stack. Install it via Homebrew (brew install aoe) or grab the install script from the GitHub repo.
Related Posts
ai-setup
Recall – Persistent Memory for Claude Code via MCP Hooks
Recall gives Claude Code a permanent memory store that survives session restarts and context compaction. Four hooks capture and restore context automatically — with cloud SaaS or self-hosted options.
2/28/2026
dev-tools
Automotive Skills Suite for AI Engineering
Evaluate Automotive Skills Suite for APQP, ASPICE, HARA, safety-plan, and DIA workflows with setup notes, governance risks, and SME review guidance.
5/28/2026
dev-tools
awesome-agentic-ai-zh Roadmap Guide
Explore awesome-agentic-ai-zh as a Chinese agentic AI learning roadmap, with setup notes, track selection, study workflow, and evaluation guidance.
5/28/2026