20 tasks. April 29, 2026 closed at 7.6x weighted leverage across 119.2 human-equivalent hours in 939 minutes of wall-clock time. Supervisory leverage came in at 115.4x.
That is 3.0 weeks of human-equivalent throughput in 15.7 hours. The ceiling was 45.0x; the floor was 3.7x. 13 of the 20 entries came from a single project.
Task Log
| # | Task | Human Est. | Claude | Sup. | Factor |
|---|---|---|---|---|---|
| 1 | auth-service: Apple Sign In server-to-server notification endpoint (verify signed JWT + handle email-disabled/enabled, consent-revoked, account-delete with session revocation and GDPR deletion path) + 14… | 6.0h | 8m | 3m | 45.0x |
| 2 | auth-service: wire Apple + Google social login through Dockerfile + buildspec.yml + populate 7 SSM params (incl. multiline .p8) + verify (391 tests green) | 3.0h | 6m | 1m | 30.0x |
| 3 | Add flashcard [content generation] stage to avian-engine (generator, writer, loop, REST endpoint, regression tests, standalone runner) | 6.0h | 22m | 8m | 16.4x |
| 4 | avian-admin: hard-delete students, login-method column, Reports tab + login-methods report (auth-service + admin-service + frontend) | 4.0h | 15m | 3m | 16.0x |
| 5 | Multi-piece overhaul: scenario 500 fix + cross-device user-state store + activity preferences (model+UI+filter) + Settings tabbed redesign + FAQ/guide content + flashcard [engine subsystem] scaffolding | 18.0h | 92m | 4m | 11.7x |
| 6 | auth-service: Apple developer-domain-association well-known endpoint + placeholder file + test | 0.5h | 3m | 1m | 10.0x |
| 7 | Debugging | 3.0h | 18m | 4m | 10.0x |
| 8 | Diagnose + fix broken avian-app-web prod CI/MCQ (root cause: empty @avian/activity-ui publishes), add Convoy cross-course UI (card + InfoButton + modal), 2 new FAQ entries, autopilot guide section, dedicated… | 9.0h | 60m | 6m | 9.0x |
| 9 | students plan/access bubbles + edit modal + bulk entitlements endpoint; engine postgres durability fix (asyncpg+ssl+ssm-loaded creds) | 5.0h | 35m | 4m | 8.6x |
| 10 | Wire failed flashcard goals into [engine subsystem] pipeline — FlashcardTribunalRepair class, prompts/schemas, write_escalated_flashcards writer, --[engine subsystem] flag in run_flashcards.py,… | 3.0h | 22m | 5m | 8.2x |
| 11 | cloud-fidelity Phase 1.a: AWS Cloudscape primitives — install + Button/Alert/StatusIndicator/KeyValuePairs wrappers + runtime dispatch shim; 290 AWS labs zero score regressions vs Phase 0 baseline | 7.0h | 55m | 1m | 7.6x |
| 12 | engine: spot->on-demand cutover (clone instance, TG swap, drain, terminate spot) + ConfirmModal replacing window.confirm | 3.0h | 25m | 2m | 7.2x |
| 13 | Fix audit_lab_instructions.py glob bug (missed 86 type-B labs); update canonical.json + 5 doc files for 2048→2134 / 935→1021 strict-pass; reconcile lab-manifest.ts with disk (added 7 missing entries:… | 3.0h | 25m | 4m | 7.2x |
| 14 | Triage + fix all 7 watch-sweep failures (3 dashboard runtime crashes; 3 initialResources schema bugs; 1 score=0 placeholder). Bring 2,134-lab corpus to strict-pass green | 6.0h | 50m | 3m | 7.2x |
| 15 | Lab post-crash recovery: cleanup + commit 1,508 in-flight changes (testId sweep + multi-checkpoint executor + sidebar nav + 90 type-B labs); fix all 9 remaining cloud-cert audit issues across AWS/GCP and add… | 4.0h | 35m | 5m | 6.9x |
| 16 | cloud-fidelity Phase 3 (GCP Material 3) + Phase 4 lite (favicon/title/LICENSING) + Modal for Azure & GCP via custom-div + vendor tokens (skip vendor Dialog entirely); 304 GCP + 427 Azure + 290 AWS = 1021… | 14.0h | 145m | 1m | 5.8x |
| 17 | Documentation | 0.8h | 8m | 3m | 5.6x |
| 18 | cloud-fidelity Phase 1.b: AWS Cloudscape Modal + LabRunner z-index 50→9000 (above Cloudscape 5000) + force-unmount on visible=false; root-caused via Cloudscape display:none vs lab autoplay waitForGone… | 6.0h | 65m | 1m | 5.5x |
| 19 | cloud-fidelity Phase 0: primitive shim + cloud detection + testid baseline + bulk import refactor (267 view files); full 2,134-lab Watch sweep regression-free; fixed pre-existing bedrock test mismatch;… | 10.0h | 120m | 2m | 5.0x |
| 20 | cloud-fidelity Phase 2: Azure Fluent UI v9 — install + Button/Alert/StatusIndicator/KeyValuePairs wrappers + FluentProvider mount + runtime dispatch; Modal deferred (Fluent Dialog blocked autoplay DOM… | 8.0h | 130m | 1m | 3.7x |
Aggregate Statistics
| Metric | Value |
|---|---|
| Total tasks | 20 |
| Total human-equivalent hours | 119.2 |
| Total Claude minutes | 939 |
| Total supervisory minutes | 62 |
| Total tokens | 2,687,000 |
| Weighted average leverage factor | 7.6x |
| Weighted average supervisory leverage factor | 115.4x |
| Human-equivalent weeks | 3.0 |
Analysis
The highest factor of the day came in at 45.0x and the lowest at 3.7x, a spread of 12.2 times between the two. That is a wide range for a single day, and it usually means the day mixed mechanical work with work that needed real judgement.
The largest single entry accounted for 6.0 of the 119.2 human-equivalent hours, or 5 percent of the day. No single task dominated the total, so the weighted average is representative.
Supervisory time was 62 minutes against 939 minutes of execution, a ratio of about 1 to 15. Supervisory leverage of 115.4x is the figure I find most honest, because it measures the hours I actually spent rather than the hours a machine spent on my behalf.
Every figure here is recorded at the time the work is done rather than reconstructed afterwards. The human estimate is my own judgement and carries the uncertainty that implies; the minutes and tokens are measured. The full dataset, including this day, is available for download.