OpenAI’s Symphony is an open‑source specification that transforms an issue tracker such as Linear into a control plane for Codex agents. The system assigns every open task to an agent that runs continuously, while humans review results. By moving from interactive Codex sessions to ticket‑level orchestration, the team reduced context switching and increased landed pull requests by 500% on some teams within three weeks. Symphony’s design lets agents pull work, manage dependencies, and monitor CI, automatically rebasing, resolving conflicts, and retrying flaky checks. It also supports speculative tasks, allowing product managers and designers to file feature requests directly into the tracker and receive review packets with video walkthroughs. The approach broadens who can initiate work and lets engineers focus on higher‑level problems. However, the shift sacrifices real‑time nudging of agents, leading to occasional misaligned outputs that are used to improve guardrails and skills. The spec shows how AI‑driven orchestration can scale engineering productivity and shift human effort toward more exploratory tasks.
An open-source spec for orchestration: Symphony
OpenAI developed Symphony, an open‑source orchestration spec that turns issue trackers into control planes for Codex agents. By letting agents pull work from Linear, the system reduced human supervision, cut context switching, and increased landed pull requests by 500% on some teams. Symphony automates task creation, dependency management, CI monitoring, and even speculative work, allowing engineers to focus on higher‑level problems.
Why it matters
By shifting routine coding to autonomous agents, Symphony demonstrates how AI can scale engineering workflows, freeing human talent for complex problem solving and accelerating software delivery.
Topic dossiers OpenAI