Ask for your PR ledger across the board

Quarterly check-in moment. You want the full strength snapshot — not just one lift, all of them, ordered however makes sense. The card you'd skim before sketching the next mesocycle.

What the agent needs to hear

Behind your sentence, the agent works backwards to two pieces: which scope and how to sort. Scope can be everything you’ve trained, a movement family (« all upper body », « squat-pattern »), or a named list (« bench, squat, deadlift »). Sort defaults to heaviest first, all-time.

Common phrasings carry their own defaults. « All my PRs » means the full ledger, heaviest first. « Top 10 PRs » is a leaderboard, weight-sorted, capped. « PRs by date » swaps to recency. Silent on both pieces, you get the full ledger heaviest-first.

Scope and ordering

The full ledger is the default: one row per exercise you’ve trained, showing that lift’s current PR. Filtered views narrow it — by movement pattern, by recency (« PRs hit in the last 90 days » surfaces what’s been moving), or by an explicit list.

Ordering decides what surfaces first. Pick the one that answers the question:

• By weight — heaviest at the top. Best for « where am I strongest ».
• By date — most recent first. Best for « what’s moved lately ».
• By progression rate — fastest-climbing first. Best for « what’s working in the current block ».
• By exercise name — alphabetical. Best when scanning for a specific lift.

How this differs from checking one PR

Asking for one PR and asking for the ledger are different decisions in disguise. The single-PR check programs your next top set — you want one number to anchor today’s session. The ledger view programs your next mesocycle — you want the wide read, not the close one.

Both are valid; the scope of the question decides which to ask. Between sessions sketching tomorrow’s bench day, ask for the one lift. Between cycles deciding which lifts to push, deload, or rotate, ask for the ledger.

When the agent gets it wrong

A PR that doesn’t match memory is the most common miss. Usually it’s an exercise-name collision — the heaviest set was logged under a similarly-named lift (« that 1RM is on flat bench, not incline »). Name the right exercise and the agent re-pulls.

Missing exercises are the second pattern. The ledger only surfaces lifts with logged working sets — if you’ve trained something but it’s not appearing, the sets were likely logged under a different name. Ask the agent to search by name to confirm.

Stale PRs are the third. A 1RM from eight months ago when you’ve been training the lift weekly since usually means the agent read « heaviest set ever » but you wanted « heaviest recent set ». Narrow the window: « PRs from the last 90 days only ».

What makes the ledger worth reading

Three things decide whether this PR ledger tells you something useful: the scope matches the decision you’re about to make (mesocycle planning needs the full ledger; today’s session needs one PR), the ordering surfaces the answer to the question you’re asking (progression rate finds what’s moving; date finds what’s gone stale), and stale PRs are flagged, not hidden (a 1RM from eight months ago isn’t wrong, but it’s not the same signal as one from last week). The system reads the ledger as the strength side of your longitudinal story. Read it before each cycle, decide where to push, and let the next month of working sets rewrite the page.