Check your active streaks

Monday morning. You don't remember whether yesterday's missed workout broke the streak or whether the streak ran on the rest-day rule. You ask the agent for the current state — active streaks, longest streaks, what counts and what doesn't.

What the agent needs to hear

Behind your sentence, the agent works backwards to two pieces: which streaks and whether you want history too. The first defaults to all currently active — training, meal logging, weight logging, plan adherence — on one card. Name one and it narrows to that lane.

The second defaults to current only. Say « and the longest of all time » and the agent pulls history alongside, so each active streak reads against its peak. « Streaks? » alone returns the full active set.

What counts as a streak day

Each streak runs on its own rule, and each one moves independently — one breaking doesn’t break the others. The training streak counts a day as on when there’s a completed session, a planned rest day, or a tagged off-plan-but-logged day. A missed session with no log and no rest tag breaks it. The rest-day rule is what most users forget: a logged rest day is a streak day, not a gap.

The meal-logging streak counts any day with at least one meal log. The weight-logging streak counts any day with one weight entry. The plan-adherence streak counts a day that hits the adherence threshold — typically 70% or above, varying by plan type. Adherence is stricter than the training streak: a session that misses the prescribed volume can keep training going while breaking adherence.

How streaks read in the wider system

Streaks are a motivation signal, not a programming input. The system reads them mostly into the user-facing weekly review (« 23-day training streak, longest this year ») and into the readiness model — long streaks of consecutive hard days nudge the « ready to push » threshold higher, because diminishing returns on back-to-back heavy sessions are real.

What streaks don’t do is drive plan adjustments or session suggestions. The agent won’t change tomorrow’s prescribed work because the streak hit 30. Reading streaks is for the « am I still consistent » check, not for tactical decisions about today’s training. That call comes from readiness and session-level reads.

When the agent gets it wrong

Three misses recur. The streak broke unexpectedly — usually a day you remember as rest that was never tagged as one. Log the rest day retroactively (« mark last Tuesday as a planned rest day ») and the streak repairs itself end-to-end. The system recomputes from the corrected timeline, not from the moment you noticed.

The wrong streak surfaced — you asked about training, the card led with plan adherence because adherence broke yesterday while training kept running. Name the lane: « just the training streak ». Longest streak missing — you got the active number but no historical comparison. Add « and the longest of all time too » and the card surfaces both, side by side.

What makes the streak read worth glancing at

Three things decide whether checking streaks is useful right now: the streaks are honest (a meal-logging streak built on minimal-effort logs is technically streak-eligible but tells you nothing about adherence), broken streaks are read as breaks, not failures (a streak that broke at 45 days because life happened still trained you for 45 days — the read is the data, not the verdict), and streaks are a glance, not a goal (chasing the streak number over the underlying work is the failure mode the streak measure was built to avoid).

Pair the streak glance with the weekly review and you get a fast read on consistency without the number becoming the metric you optimise for.