OpenAI distilled its perf-investigation playbook into a 'perf-forensics' Codex skill
OpenAIDevs · x · 2026-11-20
OpenAI's dev team walks through how Performance Factory evolved: engineers used to analyze metrics and logs manually, then captured the performance team's investigation process in a shared skill called "perf-forensics" — telling Codex which tools exist, when to use them, and how to work through a slowdown. Next, investigations moved to the cloud so an alert automatically starts a Codex session. The attached illustration shows the six-stage evolution: manual work → reusable tools → engineer+Codex → engineer dispatch → systems dispatch → factory of agents.
More from coding & agent
- OpenAI: Codex regression workflow ran 1,700+ investigations, found root causes in 600+ cases — OpenAIDevs · 2026-11-20
- Inside OpenAI's agent orchestrator: task service, dispatch loop, and event subscriptions — OpenAIDevs · 2026-11-20
- OpenAI's Codex-powered 'Performance Factory' has run 1,700+ regression investigations in 3 months — OpenAIDevs · 2026-11-20
- Gradio 7 will cut its dependencies in half over supply-chain security concerns — _akhaliq · 2026-11-20
- Vercel: agents need "headless Slack" — sync major state, don't spam humans — cramforce · 2026-11-20
- Ex-OpenAI CTO: fine-tuned SLMs can beat frontier models on specific micro-tasks — ThePeterMick · 2026-11-20