OpenAI Symphony — From Supervising Agents to Managing Work
OpenAI releases Symphony, an open-source automation tool that automatically executes agents from GitHub Issues (Linear) and creates PRs.
OpenAI Symphony — From Supervising Agents to Managing Work
OpenAI has released a new open-source project called Symphony that automates the entire workflow from GitHub Issues (Linear) to autonomous agent execution and PR creation. It represents a paradigm shift from “supervising coding agents” to “managing work.”
What is Symphony?
Symphony is a long-running automation service that continuously monitors an issue tracker (Linear in this version), creates isolated workspaces for each issue, and runs a coding agent session inside each workspace.
Traditional approach:
- Developer gives instructions to an agent
- Developer supervises the agent’s work
- Developer reviews the output
With Symphony:
- Monitors Linear board for new issues
- Automatically creates isolated workspace
- Runs agent to complete the task
- Generates PR with proof of work
- Developer only manages “work flow”
Key Features
- Issue Tracker Integration: Monitors Linear for new work items automatically
- Isolated Workspaces: Creates separate working environment for each issue
- Autonomous Execution: Launches Codex in app-server mode within the workspace
- Proof of Work: Generates CI status, PR review feedback, complexity analysis, walkthrough videos
- Safe Merging: PRs are safely merged after human approval
Technical Specifications
- Implementation: Elixir/OTP (prototype)
- Workflow Definition: Uses
WORKFLOW.mdin repository for version-controlled prompts - Concurrency: Up to 10 parallel agents (configurable)
- Auto-cleanup: Automatically cleans up when issues move to terminal states (Done, Closed)
- Security: Default safe sandbox policy (workspace-write)
Relation to “Harness Engineering”
Symphony builds on OpenAI’s “Harness Engineering” approach announced the same day. For agents to work effectively, human-readable tests, specifications, and validations are prerequisites. Symphony is the next step—automating the flow of work itself.
Future Plans
OpenAI positions Symphony as a “low-key engineering preview” for testing in trusted environments. They encourage implementations in any programming language based on SPEC.md.
Upcoming:
- Multiple issue tracker support
- More flexible workflow definitions
- Enterprise-grade security enhancements
Getting Started
git clone https://github.com/openai/symphony
cd symphony/elixir
mise install
mix setup
mix build
./bin/symphony ./WORKFLOW.md
Conclusion
Symphony represents a new paradigm where AI agents are not “tools” to be used but “workers” to be managed. Developers may shift from supervising individual coding agents to managing entire work flows.
- Repository: https://github.com/openai/symphony
- Specification: https://github.com/openai/symphony/blob/main/SPEC.md
Related Articles
Claude Code v2.1.64 Released — New Remote Control Feature and Performance Improvements
Claude Code v2.1.64 brings new Remote Control subcommand, BashTool performance improvements, enhanced auto-save, and various fixes.
Claude Code v2.1.66 Released — New Slash Commands and HTTP Hooks Support
Claude Code v2.1.66 introduces bundled /simplify and /batch slash commands, HTTP Hooks support, performance improvements, and numerous memory leak fixes.
Claude Code v2.1.68 Released — Opus 4.6 Default, ultrathink Keyword Returns
Claude Code v2.1.68 released. Opus 4.6 now defaults to medium effort for Max/Team subscribers, ultrathink keyword reintroduced, Opus 4 and 4.1 removed.
Popular Articles
Cursor Automations Release - Building Always-On Agents with Triggers and Schedules
Cursor releases Automations, a new feature for building always-on agents that run based on triggers and schedules you define.
Cursor v2.7 Release - Cursor Now Available in JetBrains IDEs
Cursor v2.7 introduces Cursor support in IntelliJ IDEA, PyCharm, WebStorm, and other JetBrains IDEs through the Agent Client Protocol (ACP).
Cursor v2.6 Release - MCP Apps and Team Plugin Management
Cursor v2.6 introduces interactive UIs in agent chats, team-based private plugin sharing, and improvements to debug mode capabilities.
Latest Articles
Claude Code v2.1.72 Released - Auto-Memory, /loop Command, and Cron Scheduling
Anthropic releases Claude Code v2.1.72 with auto-memory for context saving, /copy command, /loop command, Cron scheduling tools, Voice keybinding customization, and more.
Gemini 3.1 Pro — Competing for Top Spot in Coding & Reasoning Performance
Google's Gemini 3.1 Pro, released in February 2026, achieves 80.6% on SWE-Bench Verified and 2887 Elo on LiveCodeBench, positioning it as a top-tier reasoning model alongside Claude Opus 4.6.
Claude Code v2.1.70 Released - API Error Fixes and Performance Improvements
Anthropic releases Claude Code v2.1.70 with fixes for API 400 errors with third-party gateways, Windows/WSL clipboard improvements, and significant Remote Control polling optimizations.