Frequently Asked Questions
Everything you need to know about amux.
What is amux?
amux is the open-source control plane for AI coding agents — run, monitor, and self-heal dozens of parallel Claude Code, Codex, or Gemini CLI sessions from a web dashboard or your phone. → Getting started guide
What do I need to run amux?
Python 3, tmux, and Claude Code (the claude CLI) installed and authenticated. That's it. → 5-minute install walkthrough
How many agents can I run?
As many as your machine can handle. Each agent uses ~200-400MB RAM. On a 32GB machine, 50+ agents is practical. → Scaling to 50+ agents guide
Does amux modify Claude Code?
No. amux manages Claude Code sessions externally via tmux. It parses terminal output to detect status — no hooks, patches, or modifications.
What happens when an agent crashes?
amux's self-healing watchdog detects crashes and recovers automatically. Context exhaustion triggers auto-compaction. Thinking-block corruption triggers restart with message replay. Stuck prompts get auto-answered in YOLO mode. → Self-healing configuration guide
Can I use amux from my phone?
Yes. amux is the remote control for your AI engineering team. Install the native iOS app or use the dashboard as a PWA on iOS/Android. Monitor Claude Code sessions, approve actions, and steer agents from anywhere. → Mobile management guide
How do agents coordinate?
Via REST API. Every agent gets the API reference in global memory. They can send messages to peers, claim tasks from the board, and peek at each other's output. → Agent-to-agent orchestration guide · Kanban board guide
What does it cost?
amux is free and open source (MIT). You pay only for Claude API tokens. Running 10 agents typically costs $20-100/day depending on task complexity. → Cost calculator · vs Claude Managed Agents ($0.08/session-hour)
Is it secure?
amux is local-first with no auth built in. Use Tailscale or bind to localhost. Never expose port 8822 to the internet. → Tailscale remote access guide
Can I use it with models other than Claude?
amux is specifically built for Claude Code (the claude CLI). It relies on Claude Code's specific terminal output patterns for status detection and self-healing. → amux vs Aider (Aider supports more models)
How is amux different from Claude Code Agent Teams?
Agent Teams spawns short-lived sub-agents from a parent. amux manages long-running independent sessions with a web dashboard, self-healing, and shared task board. See detailed comparison · Running 10+ agents guide
Can I self-host it?
Yes — that's the only way to run it. amux runs on your own machine or server. No cloud service, no account needed. → Compare with Anthropic's hosted option
Get started with amux
Run dozens of Claude Code agents in parallel. Python 3 + tmux. Open source.
brew install mixpeek/amux/amux # or: pipx install amux
amux register myproject --dir ~/Dev/myproject --yolo
amux start myproject
amux serve # → https://localhost:8822View on GitHub