build-bog-brainstorm

Metadata

Status	abandoned ‖ paused
Agent identity	`f51439356729d112a6c404803d88015d5b44832c6c584c62b96732b63c2b0c7e`
Created	2026-05-06T01:18:00.792081152+00:00
Started	2026-05-06T01:18:33.494118465+00:00
Tags	`bog-this-week`, `bog-deck`, `render`, `brainstorm-deck`, `eval-scheduled`
Failure reason	framing mistake — brainstorm label belonged to the off-target MANUSCRIPT_DRAFT, not the BoG deck. Original deck framing restored.

Description

GOAL: Render slides/v2/slide_NN_*.md (15 slides) into a single PDF deck via typst.

CRITICAL FRAMING — read this before doing anything else <<< This is THE BRAINSTORM DECK. Lead-author Erik Garrison verbatim: "the brainstorm quite literally with emphasis on the storm and deemphasis of brain. lol. not the manuscript the brainstorm deck."

This is NOT a polished academic Nature-companion slide set. It is a STORM of ideas that Erik throws down at Biology of Genomes — bold, exploratory, energetic, "look at THIS!" — to pitch the project to peers and provoke discussion. Stylistic implications:

Title slide should be PUNCHY, not formal. Subtitle can lean into the "brainstorm" framing (e.g. "a brainstorm — concerted evolution and unorthodox recombination of human subtelomeres").
Bullets should be SHORT, ASSERTIVE, sometimes incomplete sentences. Not academic-prose paragraphs.
Big numbers in big fonts when they hit (e.g. 230x ER threshold, 92% within community, rho=0.715 zygotene).
The thesis pull-quote on slide 15 should land like a punch — large type, no parentheticals.
Use color and contrast generously — typst polylux supports themes; pick one with visual energy (NOT a stuffy beamer-style classroom theme).
Speaker notes are where the careful prose lives; the on-slide text is the brainstorm.
When in doubt: STORM, not brain. Erik wants to provoke and energize, not lecture.

INPUTS:

slides/v2/slide_01_title.md … slide_15_concerted_evolution_thesis.md (15 markdown briefs)
slides/v2/SLIDES_v2_PLAN.md (consolidated plan + time budget + narrative arc)
slides/v2/figure_manifest.md (figure paths + READY-TO-RENDER R scripts called out)
slides/v2/coherence_check.md (cross-slide flags)
slides/20260204_Subtelomics_overview_EG.pdf (v1 deck — NOW tracked in git, available in worktrees as of commit ~now)
paper_prep/figures/ (existing figures used by slides 04/07/10/11/12)
paper_prep/figures/nj_tree_arms/nj_tree_annotated.{pdf,png} (slide 07 NJ tree)

PIPELINE:

Render the READY-TO-RENDER figures whose R scripts are embedded in slide MDs. Per figure_manifest.md: slide_03_er_callout, slide_06_clade_callouts, slide_09_clade_legend, slide_12_stage_trajectory. Save to slides/v2/. Composites for slides 07/08/10/11 are optional polish — render them only if straightforward. Log each via wg log.
Use typst (already at ~/.local/bin/typst v0.13.1 from render-manuscript-draft-4). Use polylux (https://polylux.dev/) — battle-tested presentation framework. Pick a theme with visual energy (NOT default beamer-classroom; consider polylux themes like simple-slides with a custom punchy color palette, or touying if you prefer). Working dir: slides/v2/_typst/.
Generate slides/v2/_typst/deck.typ. Mapping: slide MD ## Title → polylux slide title (BIG); ## Bullets → punchy bullet list (short!); ## Primary figure → figure include from manifest; ## Speaker notes → polylux speaker-notes block (preserved in source, NOT on rendered slide); ## Time budget → comment at top of slide block. Big numbers should be visually emphasized (text size, color callout boxes).
Compile: typst compile slides/v2/_typst/deck.typ slides/v2/BoG_2026_brainstorm.pdf. Log to slides/v2/_typst/render.log.
MANDATORY VERIFICATION (lessons from render-manuscript-draft-4 figureless-PDF disaster): a. file slides/v2/BoG_2026_brainstorm.pdf → "PDF document" b. pdfinfo → exactly 15 pages (one per slide) c. pdfimages -list | wc -l → at least 12 image XObjects (11 figure-bearing slides + tolerance) d. pdftoppm -r 100 BoG_2026_brainstorm.pdf slides/v2/_typst/page → per-page PNG screenshots committed alongside e. If image count or page count is wrong, fail hard with the SPECIFIC error and which slide is missing its figure. NEVER produce a silent figureless PDF.
wg artifact records BoG_2026_brainstorm.pdf. Single commit: "feat: BoG brainstorm deck via typst polylux".

ACCEPTANCE: BoG_2026_brainstorm.pdf exists (15 pages, ≥12 embedded images, file=PDF), per-page screenshots in slides/v2/_typst/, wg artifact recorded, commit landed.

DO NOT:

Modify any slide_NN_*.md (read-only)
Modify paper_prep/figures/ (read-only)
Build any other deck format (.pptx, .key, beamer)
Touch files outside slides/v2/
Use SBATCH (everything light: R/ggplot2 + typst compile)
Make it look like a Nature manuscript figure caption. THIS IS A BRAINSTORM. Storm. Not brain.

GOAL: Render slides/v2/slide_NN_*.md (15 slides) into a single PDF deck via typst.

>>> CRITICAL FRAMING — read this before doing anything else <<<
This is THE BRAINSTORM DECK. Lead-author Erik Garrison verbatim: "the brainstorm quite literally with emphasis on the storm and deemphasis of brain. lol. not the manuscript the brainstorm deck." 

This is NOT a polished academic Nature-companion slide set. It is a STORM of ideas that Erik throws down at Biology of Genomes — bold, exploratory, energetic, "look at THIS!" — to pitch the project to peers and provoke discussion. Stylistic implications:
  - Title slide should be PUNCHY, not formal. Subtitle can lean into the "brainstorm" framing (e.g. "a brainstorm — concerted evolution and unorthodox recombination of human subtelomeres").
  - Bullets should be SHORT, ASSERTIVE, sometimes incomplete sentences. Not academic-prose paragraphs.
  - Big numbers in big fonts when they hit (e.g. 230x ER threshold, 92% within community, rho=0.715 zygotene).
  - The thesis pull-quote on slide 15 should land like a punch — large type, no parentheticals.
  - Use color and contrast generously — typst polylux supports themes; pick one with visual energy (NOT a stuffy beamer-style classroom theme).
  - Speaker notes are where the careful prose lives; the on-slide text is the brainstorm.
  - When in doubt: STORM, not brain. Erik wants to provoke and energize, not lecture.

INPUTS:
  - slides/v2/slide_01_title.md … slide_15_concerted_evolution_thesis.md (15 markdown briefs)
  - slides/v2/SLIDES_v2_PLAN.md (consolidated plan + time budget + narrative arc)
  - slides/v2/figure_manifest.md (figure paths + READY-TO-RENDER R scripts called out)
  - slides/v2/coherence_check.md (cross-slide flags)
  - slides/20260204_Subtelomics_overview_EG.pdf (v1 deck — NOW tracked in git, available in worktrees as of commit ~now)
  - paper_prep/figures/ (existing figures used by slides 04/07/10/11/12)
  - paper_prep/figures/nj_tree_arms/nj_tree_annotated.{pdf,png} (slide 07 NJ tree)

PIPELINE:

  1. Render the READY-TO-RENDER figures whose R scripts are embedded in slide MDs. Per figure_manifest.md: slide_03_er_callout, slide_06_clade_callouts, slide_09_clade_legend, slide_12_stage_trajectory. Save to slides/v2/. Composites for slides 07/08/10/11 are optional polish — render them only if straightforward. Log each via wg log.

  2. Use typst (already at ~/.local/bin/typst v0.13.1 from render-manuscript-draft-4). Use polylux (https://polylux.dev/) — battle-tested presentation framework. Pick a theme with visual energy (NOT default beamer-classroom; consider polylux themes like simple-slides with a custom punchy color palette, or touying if you prefer). Working dir: slides/v2/_typst/.

  3. Generate slides/v2/_typst/deck.typ. Mapping: slide MD ## Title → polylux slide title (BIG); ## Bullets → punchy bullet list (short!); ## Primary figure → figure include from manifest; ## Speaker notes → polylux speaker-notes block (preserved in source, NOT on rendered slide); ## Time budget → comment at top of slide block. Big numbers should be visually emphasized (text size, color callout boxes).

  4. Compile: typst compile slides/v2/_typst/deck.typ slides/v2/BoG_2026_brainstorm.pdf. Log to slides/v2/_typst/render.log.

  5. MANDATORY VERIFICATION (lessons from render-manuscript-draft-4 figureless-PDF disaster):
       a. file slides/v2/BoG_2026_brainstorm.pdf → "PDF document"
       b. pdfinfo → exactly 15 pages (one per slide)
       c. pdfimages -list | wc -l → at least 12 image XObjects (11 figure-bearing slides + tolerance)
       d. pdftoppm -r 100 BoG_2026_brainstorm.pdf slides/v2/_typst/page → per-page PNG screenshots committed alongside
       e. If image count or page count is wrong, fail hard with the SPECIFIC error and which slide is missing its figure. NEVER produce a silent figureless PDF.

  6. wg artifact records BoG_2026_brainstorm.pdf. Single commit: "feat: BoG brainstorm deck via typst polylux".

ACCEPTANCE: BoG_2026_brainstorm.pdf exists (15 pages, ≥12 embedded images, file=PDF), per-page screenshots in slides/v2/_typst/, wg artifact recorded, commit landed.

DO NOT:
  - Modify any slide_NN_*.md (read-only)
  - Modify paper_prep/figures/ (read-only)
  - Build any other deck format (.pptx, .key, beamer)
  - Touch files outside slides/v2/
  - Use SBATCH (everything light: R/ggplot2 + typst compile)
  - Make it look like a Nature manuscript figure caption. THIS IS A BRAINSTORM. Storm. Not brain.

Depends on

done .assign-build-bog-brainstorm

Required by

abandoned .flip-build-bog-brainstorm

Log

2026-05-06T01:18:00.774423529+00:00 Task paused
2026-05-06T01:18:05.241476085+00:00 Task published
2026-05-06T01:18:29.682904756+00:00 Lightweight assignment: agent=Careful Programmer (f5143935), exec_mode=full, context_scope=task, reason=Careful Programmer (score 0.84, 105 tasks) is ideal for this multi-step rendering pipeline with strict mandatory verification (page count, image count, screenshot generation) and explicit hard-fail conditions against silent figureless PDFs.
2026-05-06T01:18:33.494122432+00:00 Spawned by coordinator --executor claude --model opus
2026-05-06T01:18:49.446812711+00:00 Starting work — will inspect prior WIP, then render figures and build deck
2026-05-06T01:19:01.047787141+00:00 Agent 'agent-812' killed — task auto-paused (use 'wg resume' to re-enable dispatch)
2026-05-06T01:19:27.268340588+00:00 Task abandoned: framing mistake — brainstorm label belonged to the off-target MANUSCRIPT_DRAFT, not the BoG deck. Original deck framing restored.