See how last week's nutrition went

Monday morning. The new week is starting and you want a verdict on the one that just ended — how close you were to the plan, what your average macros looked like, whether you actually ate what was on paper. One read, four numbers, a headline, the week summarized.

What the agent needs to hear

Behind your sentence, the agent works backwards to two pieces: which week and which view of it. The first defaults to the most recently completed one — last Monday through Sunday. The second defaults to the full review: the headline plus the four numbers underneath.

« How was last week » is enough. If you want a specific window — « the week I started the cut », « two weeks ago » — name it and the agent re-bounds the read. If you only want one number, ask for it: « just the macro number », « how was my coverage ».

What the four numbers measure

The review unpacks the week along four independent axes. Macros answers « did you hit calories and protein, phase-aware to your goal ». Coverage answers « did you actually eat the foods on the plan ». Discipline answers « how much did you improvise — planned versus added versus swapped ». Logging answers « did you log the meals the day expected of you ». Each one sits on a 0–100 scale and ships a one-line message — « cubriste todos los alimentos del plan », « revisa improvisaciones: 2 items fuera del plan » — explaining why it landed there.

The headline on top is a weighted blend of the four. The weights depend on how the plan is set up — flexible plans lean on macros, structured plans lean on coverage, the default sits between them. You set that mode when you build the plan and can change it later.

How the auto-match changes the read

The matching that powers Coverage runs every time the review is computed. You don’t have to log « from the plan » for it to register. If breakfast called for eggs, whites, spinach and almond milk, and you logged eggs, whites, spinach and almond milk in plain language, the agent figures it out — same foods, same slot, no tag needed.

The trade-off shows up in Discipline. Coverage only cares whether the planned items got eaten; a mid-afternoon snack the plan didn’t include doesn’t lower it. But Discipline reads the ratio of planned to (planned + added + swapped), so the snack lowers that. A week can score full Coverage and middling Discipline, and the right read isn’t « bad adherence » — it’s « the plan was eaten and you added on top of it ».

When the agent gets it wrong

If the headline looks lower than what you remember, ask for the breakdown. One of the four is usually doing the pulling. Low Coverage typically means the auto-match didn’t recognize a substitution — turkey for chicken, oats for the planned cereal — flag it: « turkey on Tuesday was a sub for the chicken » and Coverage recalculates. Low Discipline means a heavy improvisation week; the dimension itself is the answer. Low Logging means the week is incomplete — meals whose window passed never got logged. The fix isn’t reading harder, it’s catching up.

If the window is wrong, name it: « the full week before this one », « May 5 to May 11 ». « Last week » is the previous Mon–Sun by default.

What makes the review worth reading

Three things decide whether this weekly read tells you something useful: the window matches the week you care about (the default is fine on Monday, ambiguous mid-week), the number you read answers the question you’re asking (Macros is « what you ate », Coverage is « how on-plan you ate », they aren’t interchangeable), and the mode on your plan reflects the kind of adherence you actually want to measure. Pick the right one, read the trend across two or three weeks, and the next plan adjustment writes itself.