What automatically consumes Codex usage without doing anything?

@Noah13

The lack of transparency around token or usage consumption isn’t unique to OpenAI—it shows up across most platforms in this space.

While it’s fine to raise concerns here on the forum, it’s worth noting that OpenAI developers tend to be more responsive on GitHub, particularly in the Codex issues tracker:
→ OpenAI Codex issues: Issues · openai/codex · GitHub

A few relevant threads you may want to review:

These threads show that others are running into similar problems, and they occasionally include useful developer feedback or workarounds.

If this is something you care about, the most effective step is to upvote (:+1:) the relevant issues—especially #14593—and track them for updates. That’s currently the primary signal used to prioritize feature work.