amux vs OpenAI Symphony

Developer-driven Claude Code fleet vs autonomous Codex pipeline — same goal, fundamentally different execution model.

OpenAI Symphony (open-sourced June 2026) is a ticket-driven autonomous coding pipeline. It polls a Linear issue board, claims tickets, spawns Codex CLI agents in parallel, and opens GitHub pull requests — no human input required during execution. Written in Elixir/OTP, MIT licensed. Designed as a reference implementation for autonomous coding at OpenAI's internal scale.

amux is an open-source control plane for running a parallel AI engineering team. It runs Claude Code sessions in tmux, monitored from a web dashboard or mobile app, with a built-in kanban board, self-healing watchdog, and REST API for agent coordination. MIT + Commons Clause, single Python file, zero external dependencies.

For the deep-dive comparison including architecture breakdown, workflow analysis, fault tolerance details, and the "500% PR increase" claim, see the complete amux vs Symphony comparison.

The core difference: autonomous pipeline vs developer cockpit

Symphony's design is fully autonomous: connect it to Linear and Codex, and it runs indefinitely without human intervention. Tasks flow from issue to PR with no dashboard to monitor and no controls to steer mid-run. This is powerful for teams with well-defined, tightly scoped tickets and a high degree of trust in Codex's output.

amux gives you a cockpit. You see every session, its current status, token spend, and live output. You can steer a running agent, replay a failed session, or pause the board. The self-healing watchdog handles failures automatically, but you're always able to intervene. This works better for codebases where human judgment is still part of the loop — most production systems.

Feature comparison

FeatureamuxOpenAI Symphony
Agent modelClaude Code (Anthropic)Codex CLI (OpenAI)
Execution modelDeveloper-driven dashboard + queueFully autonomous pipeline
Issue trackerBuilt-in SQLite kanban boardLinear (only, at launch)
Web dashboardFull UI — sessions, board, peek, spendNone in reference implementation
Mobile managementNative iOS app + PWANone
Self-healingPython watchdog — context overflow, crash, stuck, rate-limitElixir/OTP supervision trees
Fault toleranceAuto-compact, restart with message replay, YOLO modeProcess supervision + exponential backoff
RuntimePython 3 + tmuxElixir/OTP (BEAM VM)
Setup complexityLow — ./install.sh, 5 minHigh — Elixir, mise, Linear + Codex API keys
Human-in-the-loopOptional (YOLO mode for full autonomy)Not in reference implementation
LicenseMIT + Commons ClauseMIT
MaintenanceActive — iOS app, cloud deploy, regular commitsReference implementation, not a maintained product

When Symphony is a better fit

When amux is a better fit

See also

Get started with amux

Run an AI engineering team from your dashboard or phone. Open source, Python 3 + tmux.

git clone https://github.com/mixpeek/amux && cd amux && ./install.sh
amux register myproject --dir ~/Dev/myproject --yolo
amux start myproject
amux serve  # → https://localhost:8822
View on GitHub