How build activity is measured¶
Activity shows two numbers for a period: build time, how long you were actively building, and the flow score, how smoothly your build loop ran. Both come from what your tools report as you work: the CLI, the TUI, your coding agents and your editor. There are no timers to start and no surveys to fill in. This page explains what each number counts and how it is worked out. For where to find them, see Home, Activity and other screens.
Build time¶
Build time is the time you were actively building, not the time your tools were open. A session counts only while it shows activity; a session left open while you were away has its idle time trimmed.
When several sessions run at once, say you and two agents working side by side, that time is counted once. Build time is the union of your active time across all your sessions, so an hour with three agents busy is one hour of build time.
Session time is the plain sum of every session's active time, overlaps included. It is never less than build time; the gap is time when more than one session was active.
Agent orchestration is session time divided by build time. 1× means one active session at a time. 3× means that, on average, three sessions were working at once while you were building. It is a measure of how much work you ran in parallel, not of productivity, and is best read beside the flow score and what cost it.
The receipt under build time lists each day (build time against session time) and each session: its tool, its wall time from start to end, the time that counted, and the idle time trimmed.
The flow score¶
The flow score is a number from 0 to 100 for how well the platform kept you in flow: 100 is long, uninterrupted stretches of work with fast feedback and nothing in the way; 0 is a day of failures and retries. It measures the build environment, not your output. Every point lost is traced to something that happened, so the score also says what to fix.
Four parts¶
Each day is scored on four parts, each from 0 to 100:
| Part | Measures | Today's weight |
|---|---|---|
| Focus | The share of your active time spent in uninterrupted runs long enough to reach flow: at least 20 minutes, with no silence longer than 15 minutes. | 35% |
| Feedback | How quickly define, build and deploy finished against their targets. A fast failure is still fast feedback; failures cost points under Friction instead. | 25% |
| Friction | Starts at 100 and loses points for failed operations, retry churn (several runs on the same service inside an hour, most of them failing) and skill searches that found nothing. Each kind of loss is capped, so one bad loop can't zero a day. | 30% |
| Progress | Forward motion through define, build and deploy: a green deploy counts for more than a green build, and a build for more than a define. It carries the smallest weight on purpose. | 10% |
A part with nothing to measure on a day, Feedback on a day with no builds, for example, is left out, and the other parts' weights are shared out among the rest. Missing data never lowers a score.
Days, and the period¶
A day is scored once you have at least 15 minutes of active time on it. A day with less has no score, not a score of 0: no activity means nothing to rate.
The period's score is the average of its scored days, with each day counted by how much active time it had, so a full day of work weighs more than a short one.
What cost flow¶
Under the score, What cost flow lists the things that took points away, largest first, such as "3 failed builds" or "retry churn (5 runs, 3 failed)", with the points each cost and where it happened. These are the clearest places to start improving your build loop.