Engineering teams already using coding agents such as Codex, ideally with a codebase set up for harness engineering, who are comfortable running experimental software in a trusted environment.

What can you build with symphony?

Turn tickets on a Linear board into pull requests without sitting beside a coding agent. Build your own Symphony in your preferred language by giving SPEC.md to a coding agent. Run the Elixir reference implementation against a repository to try board-driven agent workflows. Review agent output through CI results, walkthrough videos and complexity analysis instead of reading every step.

openai /

symphony

Watches a work board and hands each task to its own isolated coding agent run

27.8k stars
288 stars since 3 Oct
Elixir
Updated 37 days ago
Advanced
Coding agents & dev tools

Open on GitHub (opens in a new tab) Website (opens in a new tab)

Symphony is from OpenAI. It moves you from watching a coding agent work to managing the work itself. You put tasks on a board, and Symphony makes sure agents pick them up and finish them.

In the demo, Symphony monitors a Linear board for new work and starts an agent for each task, each in its own isolated run. The agents report back with proof of work: CI status, PR review feedback, complexity analysis and walkthrough videos. When a person accepts the result, the agents land the PR.

The repo gives you two routes. The first is a written spec (SPEC.md) that you can hand to a coding agent to build Symphony in any language. The second is an experimental reference implementation in Elixir. The README says it works best in codebases that already follow OpenAI's harness engineering approach, and it calls the project a low-key engineering preview for trusted environments.

What it does

Board-driven task intake
Symphony monitors a work board, with Linear shown in the demo, and spawns agents to handle the tasks it finds.
Isolated implementation runs
Each piece of work gets its own autonomous run, so agents do not need someone watching them.
Proof of work
Agents report CI status, PR review feedback, complexity analysis and walkthrough videos so a person can judge the result.
Landing PRs
Once work is accepted, the agents land the pull request.
Spec plus reference implementation
SPEC.md describes the system so you can build your own version. An experimental Elixir implementation is included.

What you could build

Turn tickets on a Linear board into pull requests without sitting beside a coding agent.
Build your own Symphony in your preferred language by giving SPEC.md to a coding agent.
Run the Elixir reference implementation against a repository to try board-driven agent workflows.
Review agent output through CI results, walkthrough videos and complexity analysis instead of reading every step.

Get started

Read elixir/README.md to set up and run the Elixir reference implementation, or ask your coding agent to set it up for your repository from that file. To build your own version, give your coding agent the SPEC.md link.

Commands are copied from the project's README. Read it on GitHub for the full setup.