Holding Area

The premise

Some items don’t deserve your attention even once. Marketing emails. Automated notifications you glance at and dismiss. “FYI” updates that don’t require a response. The cost of having them in your loop isn’t reading them — it’s having to act on them, even if “acting” means pressing e to dismiss.

The Holding Area is a side queue for those items. The AI summarizer classifies new items 1–5 on urgency. When you have the Holding Area enabled and the AI rates an item 1 (irrelevant), the item goes into the Holding Area instead of your loop.

You triage the Holding Area in batches — at the times you choose — not in real time.

How items get there

Only the AI summarizer pipeline routes items to the Holding Area. That means:

• Gmail, GitHub, and Linear stream items the summarizer rates as urgency 1 → Holding Area.
• Items you create manually → never Holding Area, regardless of urgency.
• Items from streams that don’t run the summarizer → never Holding Area.
• Manually changing an item’s urgency to 1 with ! → does not move it to the Holding Area. The Holding Area is the AI’s filter, not yours.

This keeps things predictable: only the summarizer’s “this doesn’t matter” verdict on incoming items causes hiding.

Reviewing the Holding Area

Two ways to get there:

1. The Holding Area NavTile in your Preferences page (visible only when the feature is on).
2. A “Review your Holding Area (N items)” item that lands in your loop at the times you’ve configured. Pressing Enter on that item takes you to /holding. The reminder only appears when there’s actually something to review.

In the Holding Area:

• j / k to move
• e (or Enter / x) to complete the item — confirming it was irrelevant.
• ! to send the item to your loop. Sending sets urgency to 2 and gives the item a fresh slot at the top of your loop so you’ll see it next.
• o to open the item’s URL in a new tab.
• Esc to leave.

The five urgency levels and their colors (also visible in your loop):

Configuring it

In Preferences → Holding Area:

• A toggle to enable the feature (default off).
• A list of times in your local timezone for review reminders. Add times like 09:00, 14:00, 17:00. Empty list = no auto reminders; you can still visit /holding whenever you like.

Switching the feature on does not retroactively classify items already in your loop. Only newly-summarized items going forward will be routed.

Feedback loop with lookbacks

When you complete an item rated urgency 1 that was not in the Holding Area, your daily lookback notices. Either you didn’t have the feature on, or the AI hadn’t rated it as 1 — both signals that something the AI should have caught slipped through. The lookback’s context_suggestion machinery uses that bucket to propose context lines to add to your AI context, like:

“Newsletters from *@substack.com are always urgency 1 for me.”

“Order-shipped notifications from retailers don’t need to land in my loop — treat them as urgency 1.”

When you accept the suggestion, that context flows into every future summarization, so similar items go straight to Holding Area.

When the AI gets it wrong

If you find a held item that did matter, hit ! to send it to your loop. The Holding Area is a high-recall filter; it’s tuned to be willing to set urgency 1, on the assumption that you’ll occasionally rescue something. The lookback feedback loop is the slow path that improves accuracy over time.

If you’ve held an item you’d rather not have held, you can also amend your AI context directly to include “items like X should be at least urgency 2” — that gets baked into the next summarization run.