OpenAI Symphony: SPEC.md for Autonomous Coding Agents

OpenAI has open-sourced Symphony, a SPEC.md that describes how to orchestrate autonomous coding agents. This system uses project management tools, like issue trackers, as a control plan. Here's how it works: Symphony coordinates multiple coding agents by assigning tasks to dedicated agents. These agents work autonomously until their task is complete. A human then reviews the output. OpenAI engineers created Symphony to overcome a "human attention" bottleneck. Previously, engineers managed a few coding sessions at a time, but context switching became difficult. Symphony changes this by structuring its workflow around project deliverables like issues, tasks, and milestones, rather than individual coding sessions. Symphony continuously monitors the task board, ensuring every active task has an agent running until it's done. If an agent crashes or stalls, Symphony restarts it. If new work appears, Symphony organizes it. This approach significantly reduces the cost of agent mistakes, as it primarily involves reviewing and potentially rejecting completed work. What's interesting is that Symphony is not a complex supervision system itself. Instead, it's a SPEC.md file. This means any organization can use it to create their own orchestrator. The reference implementation uses Elixir. The bottom line is that OpenAI doesn't position Symphony as a standalone product. It's a reference implementation for developers to adapt, offering a new way to manage autonomous coding.