Read the explanation
The daily operations preset lists twelve support events, one summary and four draft pull requests, seventeen tasks per day. Fourteen simulated days give two hundred thirty eight tasks. Offload is forty plus forty two times trust, so trust point seven five gives seventy one point five percent. Interventions round total tasks times one minus point seven five times trust, times point three five, yielding thirty six. Weekly hours saved assumes fifteen minutes for every task and multiplies by offload, giving twenty one point three. Habit score rounds ninety minus point two times interventions, or eighty three. Bars compare the stored initial offload sixty eight point four with formula seventy one point five at two pixels per caption point. These are authored assumptions, not measured productivity or behavioral evidence. The native trust slider ranges from point one to point nine five in steps of point zero five. Raising it to its maximum point nine five increases offload to seventy nine point nine percent and lowers the intervention caption to twenty four for this fixture. Saved hours becomes twenty three point eight per week under the same fifteen minute assumption. The habit score falls because trust above point nine subtracts fifteen from its base, taking ninety to seventy five. Subtracting point two times twenty four rounds seventy, compared with eighty three at default trust. Bars span one hundred sixty six and one hundred forty at two pixels per habit point. This discontinuity is an authored threshold, not evidence of actual user behavior. Delegation confidence thresholds are configuration text rather than a running model deciding tasks. The trigger export includes stated daily frequency, but its cron expression maps every frequency other than one to every thirty minutes. A stated twelve daily events and four daily events therefore both export a nominal forty eight daily schedule if interpreted continuously. Bars compare twelve versus forty eight at four pixels per daily event. Initial DOM ready attempts D3 graph setup before computing telemetry, so missing D3 leaves initial captions. Trust listeners were bound earlier and can still compute telemetry and generate export text. The run simulation advances a caption every one hundred fifty milliseconds and animates packets; it does not execute fourteen days of real tasks. Actual local exports and preserved original geometry are separate evidence, without scheduler activation, empirical outcomes, exhaustive functions or public deployment.