Position Memory

Position Memory is the toggle on the Tap editor that lets Mavster reuse a remembered tap position when the screen still matches, instead of resolving the target from scratch. It's the highest-leverage cost saving Mavster offers for stable tests.

What the toggle says

The verbatim label and description in the editor:

Position Memory

"Remember and reuse element position across runs. Only enable for elements with stable, fixed positions (e.g., navigation buttons). Automatically validates before use."

On by default for Tap. When memory has been stored, a Clear button appears under the toggle to wipe it and force a fresh resolution next run.

When to leave it on

The toggle's own description is the right rule of thumb: enable for elements with stable, fixed positions — navigation icons, tab bar items, persistent buttons that don't move between runs. The hit rate is high and the validation cost is low.

When to turn it off

• Targets that move between runs. Cards in a feed that re-order, items in a sortable list, anything driven by content the test doesn't control.
• Targets behind dynamic layouts. A button whose position depends on whether a banner is visible.
• The first time you debug a flaky step. Disabling Position Memory tells you whether the flake comes from stale memory or from the underlying AI search; turn it back on once you've ruled stale memory out.

Local to each Mac

Position Memory is local to the Mac that produced it — it doesn't follow your project across machines. A new Mac running the same test takes a couple of warmup runs before it reaches its fastest, cached steady state.

Cost framing

A memory hit is much cheaper than resolving a tap from scratch — not free, but a small fraction of the cold-start cost. Multiplied across a test that runs daily on dozens of stable taps, the saving adds up fast. Leaving it on for an unstable target costs roughly the same as having had it off, so there's almost no downside even when in doubt.