Product feedback: “Confetti cannon” in Codex settings sends the wrong signal about priorities

Product feedback

I was genuinely surprised to find a dedicated “Confetti cannon” toggle in Codex settings.

A playful feature like this would be completely normal as an easter egg or something a user adds to their own setup for fun. What surprised me is seeing it elevated into the settings of a corporate engineering product used at large scale.

The casual user asking ChatGPT a simple everyday question is unlikely to ever reach this part of the settings. The people who do reach it are much more likely to be looking for serious controls over agents, permissions, execution, sandboxing, approvals, session state, usage visibility, reliability, and workflow behavior.

So the question is not really “why is there confetti?” The question is: what does its visibility say about product priorities?

There is also a resource-allocation concern. OpenAI regularly talks about the cost and scarcity of compute, usage limits, and the expense of advanced agentic workloads. As a paying user, that makes prioritization matter. I am not claiming that this feature consumed some enormous amount of compute, but engineering, design, testing, product attention, and potentially model/tooling resources were still spent on it while more important operational needs remain unresolved.

I am not asking OpenAI to remove fun from the product. I am asking whether a novelty feature like this belongs at the same level of visibility as controls that materially affect real engineering workflows.

Please treat this as product/UX feedback, not a bug report.

Thread ID from Codex feedback flow: 01a0759d-13af-76f2-ad56-4bad236a5a07

The issue names no files, tests, or entry points. Start by reviewing the Codex settings UI and the referenced feedback thread; a decision about the visibility and placement of the “Confetti cannon” toggle would define what done means.