Skip to main content
AI02 September 2026

Coding and testing

18 tasks. September 2, 2026 closed at 11.3x aggregate execution leverage across 665.5 qualified-senior-equivalent hours in 3,547 agent-session minutes. Covered operator leverage was 314.4x.

That is 16.6 qualified-senior-equivalent weeks. Task factors ranged from 4.4x to 39.3x. The largest project grouping contained 12 entries.

Task Log

#TaskSenior Est.Agent SessionOperatorFactor
1[product] advanced-math activity package extraction, Phases 0-3 of the 2026-09-02 plan. Phase 0: [engine subsystem] across 5 repos; boundary frozen by source analysis of production imports (verification +…36.0h55m2m39.3x
2Testing2.5h4m3m37.5x
3Deployment40.0h72m2m33.3x
4Testing4.5h10m3m27.0x
5Deployment22.0h57m7m23.2x
6[product] /findings run + 0.9.0 release: verified, ranked and resolved 15 findings against the 0.8.0 math pipeline (13 from the owner, 2 raised during the run). Each fix shipped with a regression test…41.0h139m13m17.7x
7Deployment130.0h461m12m16.9x
8Testing12.5h50m6m15.0x
9notification-service /findings run on the Fable review of the notif-identity-durability program. Took 6 reported findings (3xP1 + 3xP2), verified every one before fixing any (three parallel read-only agents,…72.0h290m3m14.9x
10[product] (@[product]/activity-ui) 2.9.3 — eighteenth-pass symbolic verification remediation. Verified 4 owner-reported findings (V72-V75) against the real pinned mathjs 15.2.0 and the full 24-bank /…24.0h97m4m14.8x
11[product] advanced-math extraction, Phase 4 — web bridge in [product]. Composed @[product]/math-activity-ui@1.0.0 over the core registry through the new registries property, with replace for the eight…10.0h42m1m14.3x
12/findings run on @[product]/subscribe-react: verified and resolved an 11-finding owner review of the embedded Stripe checkout, plus 4 defects found en route. Six were money-or-identity S0s: recovery being…50.0h219m30m13.7x
13Testing24.0h111m5m13.0x
14Testing22.0h128m5m10.3x
15Testing32.0h240m6m8.0x
16Closing Round ([product] math pipeline, findings round Q). Orchestrated a 7-of-9-phase program fixing 9 of 10 findings in the LessonView math pipeline, building a property-based oracle tier graded against the…55.0h440m6m7.5x
17Infrastructure44.0h528m6m5.0x
18Testing44.0h604m13m4.4x

Aggregate Statistics

MetricValue
Total tasks18
Qualified-senior-equivalent hours665.5
Agent-session minutes3,547
Covered operator minutes127
Recorded tokens1,917,405
Aggregate execution leverage11.3x
Covered operator leverage314.4x

Interpretation

The qualified-senior hours are counterfactual estimates and inherit that uncertainty. Execution, token, and operator-time values retain their recorded provenance, which is incomplete for some legacy rows. Summed agent-session minutes are not serial wall-clock time when sessions overlap.

This is one unusually experienced operator's record under unusual working conditions, not a general benchmark. The complete reconciled snapshot is available for download.