End a program when the block is done

Twelve weeks of the strength + cut program done. You hit the milestones, the next block starts Monday. Before flipping into the new program, you close this one out so the summary captures the whole arc — adherence, PR landings, body delta, weekly volume curve.

What the agent needs to hear

Behind your sentence, the agent works backwards to two pieces: confirmation that you’re ending the program and, optionally, a closing note — « hit the target », « calling it early — too aggressive », « switching to the next phase ». The active program is implicit; only one runs at a time, and the agent already knows which.

Before anything flips, the agent renders the closure summary on a preview card — adherence across both routine and diet plan, PRs landed during the program, body composition delta, weekly volume trajectory — and waits for a nod. If the arc reads honestly, « yes, close it » settles it.

What gets captured at program close

The closure freezes a handful of metrics against the program window. The training adherence is computed across every scheduled session from start to close. The nutrition adherence rolls up the linked diet plan’s daily targets across the same window, if a plan was bundled in. The PRs landed count surfaces with which lifts moved — bench, squat, deadlift, whichever broke during the block.

The body delta reads start weight and key perimeters against end values. The average weekly volume captures the trajectory — ascending, plateauing, deloading. The session count vs planned flags how much you actually executed. And the closing note, if you offered one, attaches as the last word.

Once saved, the summary is immutable. The program moves to history frozen as of close. Past sessions and meal logs stay intact and readable — nothing about them changes.

Ending a program vs ending its components separately

A program bundles a routine engagement and a diet engagement. Ending the program closes both engagements together and bounds them in a single « this block » read. Adherence, volume, and body delta resolve against the same window.

Ending only the routine — without closing the program — leaves it technically active with a half-closed state. Same the other way around: ending only the diet engagement corrupts the bounded read. The program-level numbers stop reconciling. The forward-looking rule still holds: ending the program doesn’t rewrite past sessions or logs. It freezes them as the program’s history and lets you activate the next program cleanly.

When the agent gets it wrong

Catch the wrong move on the preview card before you nod. The agent ended the wrong program: « no, the strength + cut — the one running now, not the maintenance from last quarter ». The card swaps.

The agent only closed the routine without ending the diet engagement: « close both — they were bundled ». The card redraws with both layers wrapping in the same beat. The closure date is wrong: « I finished Friday, not today — backdate the close so the program window reads accurately ». The window shifts and the adherence recomputes against the corrected dates.

What makes a clean program close

Three things decide whether this program closes usefully: the closing summary captures the program’s actual arc (a 12-week program closed a week late stretches the adherence window; closed early truncates it), both linked engagements close together (a half-closed program corrupts the bounded read that programs exist to give you), and the closing note carries the intent (« hit the target », « bailed at week 8 — wrist injury », « solid block, ready for the next » — all valid signals for designing the next program). A clean closure makes the next program’s targets more informed — the history is what the system reads when you ask what worked.