OpenAI releases Symphony, an open-source spec for agent orchestration that turns a project-management board like Linear into a control plane for coding agents
Six months ago, while working on an internal productivity tool, our team made a controversial (at the time) decision: we'd build our repo with no human-written code.
OpenAI
Context & Ripple Effects
OpenAI’s agent tooling has progressed from APIs and an orchestration SDK to sandboxing and evaluation support for longer-running tasks. Symphony extends that stack into the work-management layer, using an existing project board as the interface for coordinating coding agents.
The release also arrives amid competing efforts to open-source coding agents and establish shared agent standards. Its significance is less a new model than an attempt to standardize how agents are assigned, tracked, and governed inside software-development workflows.
First-order effects
Teams using Linear-like boards can use Symphony’s open specification to connect task management directly to coding-agent orchestration, making the board a place to dispatch and monitor agent work.
OpenAI broadens its developer-agent offering from building and testing agents to defining a workflow layer around them; Linear becomes a relevant integration point rather than merely a planning tool.
Second-order effects
Coding-agent vendors and project-management platforms face pressure to support interoperable task, status, and handoff conventions rather than require users to operate separate agent consoles.
Organizations can centralize agent work around existing project controls, increasing the importance of permissions, sandboxes, and evaluation tooling as agents take on longer-running coding tasks.
Third-order effects
If adopted across tools, open orchestration specifications could shift differentiation from basic agent dispatch toward reliability, governance, integrations, and the underlying coding models.
Software-development work surfaces may increasingly serve as operational control planes for mixed human-and-agent teams, though that outcome depends on whether specifications such as Symphony gain support beyond their originating ecosystem.
The trend: Agent infrastructure is moving from standalone assistants toward workflow-native, governed control planes that coordinate multiple agents inside the tools where work is already planned and reviewed.
📣 What if every open issue had a Codex agent? That's the idea behind Symphony, an open-source agent orchestrator for Codex that turns task trackers into always-on systems for agentic work, letting humans focus on review and direction. [video]
Engineers at OpenAI experience the same problem as everyone else — we can supervise about 3-5 coding agents. After that productivity drops. Codex is smart, but our attention is limited. So we built (and open sourced!) Symphony to remove that ceiling. Here's how it works:
We quietly launched the Symphony repo on Github last month, and it's already accumulated 15.5k stars! Excited to share this post that dives into it even deeper: a library that lets you use Codex to orchestrate work normally done by teams of engineers. https://openai.com/...
Symphony is amazing. Best agent orchestrator I've used. It's open-source and you can use it with your ChatGPT sub. Only downside I found: it's *VERY* token hungry. Closed 30 Linear issues with Symphony + Codex in a week and it was the closest I came to my weekly Codex limit.
Symphony: a minimal orchestration layer for Codex It connects issue trackers like Linear to coding agents, so each task can spin up its own Codex session In a nutshell: open issue → assign agent → generate PR → human review https://openai.com/...
Learning how to use coding agents effectively is the most interesting engineering problem in the world right now. The solution @alex_frantic came up with for our team is Symphony. I think Symphony has a few really interesting ideas embedded in it: 1. The approach itself.
We have reached a moment where instead of releasing software you simply release the detailed spec for software and tell people to prompt their agent to build it themselves — From the README of OpenAI's new Symphony orchestrator: github.com/openai/symph... [image]
free idea for someone: verifiable build pipelines that use specific models to “compile” specifications like github.com/openai/symph... into code, then compile that code into binaries. — I care to audit the spec, I care not to audit each line of code, and I just want to know bin…