Keep the context that earns its place.

Inventory your AI setup, expose overlap, and leave with a reversible seven-day experiment. Token counts are planning estimates, never provider measurements.

A visual reference from the original discussion about simplifying Claude setups

Inspired by a simple claim: productive systems can get better by removing what no longer helps.

Your context inventory

Describe structure, not private content.

Items

The lean plan

Actions will appear here.

Back up before removing

Take the decision with you

Save a checkpoint or export the assumptions behind it.

Seven-day experiment

Apply one reversible change, repeat one real task, and compare quality before changing anything else.

Add context item

Reset the current audit?

The representative sample will return. Saved scenarios stay available.

Enjoy this tool? Build your own with Super