Editorial illustration for Open-source orchestration spec 'Symphony' invites agents to build implementations
Symphony: Open-Source AI Agent Orchestration Spec
Open-source orchestration spec 'Symphony' invites agents to build implementations
Symphony began as a single Codex session running in tmux, polling Linear and spawning sub-agents for new tasks. From that scrappy seed, its second iteration embedded itself into our main project repository, a repo already built with agents in mind. We had the harness, the skills, the context.
Symphony simply connected it all. Once the basic functionality held, we turned the system on itself: Symphony built Symphony. The reaction when we demoed agents managing tasks and attaching proof-of-work videos?
Electric. The project channel swelled. Teams adopted it without being told.
That internal product-market fit wasn’t accidental; it was the precondition for launching externally. Now we’re open-sourcing the spec, and we’re inviting your favorite coding agent to write its own implementation.
Symphony(opens in a new window) is an agent orchestrator that turns a project-management board like Linear into a control plane for coding agents.
Symphony is not a product. It is a provocation. A spec that dares agents to build their own implementations, to iterate, to compete, and to collaborate.
We gave it space to evolve from a single tmux session into the connective tissue of an entire engineering organization. Now we give it to the world. The code is the argument.
The implementations will be the proof. What emerges next is not ours to predict, it is yours to build.
Common Questions Answered
How does the Symphony orchestration spec enable automated code generation?
Symphony provides a minimal open-source specification that allows coding agents to autonomously generate and coordinate software implementations. By defining a framework for agents to poll task trackers, spawn sub-agents, and work collaboratively, Symphony aims to reduce human intervention in software development processes.
What was the initial implementation of the Symphony orchestration spec?
The first version of Symphony was a Codex session running in tmux that continuously polled Linear for tasks and spawned sub-agents to handle those tasks. The second version was integrated directly into the main project repository, which already had an agent harness designed to provide agents with necessary skills and context.
What is the primary goal of making the Symphony spec open-source?
By publicly exposing the Symphony specification, its creators are challenging the coding agent community to develop their own implementations of the orchestration framework. The open-source approach encourages experimentation and collaborative development of automated software generation techniques.
Further Reading
- OpenAI Releases Symphony: An Open Source Agentic Framework for Orchestrating Autonomous AI Agents through Structured Implementation Runs — MarkTechPost
- OpenAI Symphony: When AI Agents Run Your Sprint Board — SJ Ramblings
- Introduction - Symphony — Mintlify
- Symphony: A Decentralized Multi-Agent Framework for Scalable Orchestration — arXiv