Artfical AI / Docs
tCode overview Open tAI Install tCode
How it works

A reasonable habit for when to reach for it

A reasonable default habit is reaching for plan mode on anything you are not already fully certain about the shape of, and skipping it for anything you already know exactly how to describe in one clear request.

There is essentially no cost to using it more often than strictly necessary, beyond the small extra step of reviewing the assembled plan before approving it to actually run.

For a change you were already planning to review carefully either way, that extra step is not really extra work at all, it just moves the review earlier, before the edits exist, rather than after.

A useful rule of thumb: if you find yourself wanting to see the whole shape of a change before committing to the first piece of it, that instinct is itself the signal plan mode exists to answer; there is rarely a downside to trusting it.

Over time, most people settle into a fairly consistent personal threshold for when they reach for it, without needing to consciously re-derive the decision on every single request; it becomes as automatic as deciding whether a task needs a search first or not.