Plan — Repo restructure

2026-refresh › repo-restructure · chore · 2026-08-05 · spec.html

Plan — Lecture production

2026-refresh › lecture-production · feature · 2026-08-05 · spec.html

Three phases: settle the definitive recording list · prove the recording setup end to end with a real test take · then write every script + workshop pair, section by section, with Asim recording at his own pace as material becomes ready. Recording itself is Asim-paced and tracked, not phased.

F1

Reconcile the definitive recording list

The production sheet's deprecated tier framing resolved into one flat list: every lecture to record, its section, target length, and script order — confirmed by Asim.

Proved byThe recording sheet (HTML): every new/re-record lecture listed with section, target duration and status columns — and Asim's confirmation of the list, including the S16/S17 keep-or-trim call.
the work + evidence drawers

Build

  • diffRecording sheet built from triage + production sheet + the August pivots (~24 new + 4 re-records + on-camera intro; Map/Set re-record included)
    evidence
    On tick.

Prove

  • logAsim confirms the list (recorded)
    log
    On tick.
F2

Prove the recording setup with a real test take

Recording profile finalised (Code Runner F8 on markdown, layout, theme per design-system), capture app tested (two .mov files), mic checked through the loudness pipeline, takes-inbox convention created.

Proved byA ~60s test take of the let-const specimen (face.mov + screen.mov in the inbox): F8 runs a block on camera, both files conform after normalise, mic loudness lands in range — and Asim says the setup felt effortless.
the work + evidence drawers

Build

  • diffRecording profile: Code Runner + F8 keybinding + markdown executorMap settings + layout (file left, output right) + theme/font from design-system; one-page setup checklist
    evidence
    On tick.
  • logCapture test with Asim: chosen app records face.mov + screen.mov; files probe clean (resolution, fps, audio present)
    log — ffprobe both files
    On tick.
  • diffTakes inbox convention: courses/adv-js/media/source/takes/<slug>/ documented in the checklist
    evidence
    On tick.

Prove

  • logThe 60s test take: both files in the inbox, conform-check on the normalised pair, Asim's verdict on the feel
    log
    On tick.
F3

Write every script and workshop file, section by section

The specimen format stamped across the list: verbatim say/do script + pre-built workshop .md per lecture, all code executed on Node before scripting; delivered in section batches so Asim can start recording early.

Proved byThe recording sheet fully green on the script side: every listed lecture has its script + workshop pair; a runner log shows every workshop block executing clean; Asim has signed each section batch as "I could record these".
the work + evidence drawers

Build

  • diffScripts + workshop files for the priority sections first (S4 scopes gap, S16 Interview Lab, event-loop re-records), then the rest — specimen format, traps lists, continuity refs
    evidence
    On tick.
  • logEvery code block in every workshop file executed on Node LTS; outputs quoted verbatim in scripts; a repeatable runner covers the whole examples tree
    log — runner output
    On tick.
  • diffRecording sheet tracks per-lecture state (scripted → recorded → in-editing) as takes arrive
    evidence
    On tick.

Prove

  • logSheet audit: zero listed lectures without a script+workshop pair; runner log clean; Asim's per-section sign-offs recorded
    log
    On tick.

Decision Defaults — standing autonomy for this run

  • Script wording, example choice and beat structure are the writer's call within the specimen format and traps discipline — Asim redlines at section sign-off.
  • Section batch order may flex if a dependency (design theme, Asim availability) suggests it — logged.
  • Not covered (always a stop): adding/removing lectures from the confirmed list · changing the recording contract (two .mov files) · anything requiring Asim's mic beyond the list · recording a verdict without a verifier's report · amending this plan.

Decision Log

Empty — fills during the run.