qa-quarto

star 7

Adversarial QA comparing Quarto slides to Beamer benchmark. Run critic (find issues) then fixer (apply fixes); repeat until approved or max 5 rounds. Use when the user asks to QA Quarto vs Beamer or validate a lecture translation.

tesolchina By tesolchina schedule Updated 2/21/2026

name: qa-quarto description: Adversarial QA comparing Quarto slides to Beamer benchmark. Run critic (find issues) then fixer (apply fixes); repeat until approved or max 5 rounds. Use when the user asks to QA Quarto vs Beamer or validate a lecture translation.

Adversarial Quarto vs Beamer QA

Beamer PDF is the gold standard. Quarto must match or exceed it.

Hard gates (non-negotiable)

  • No content cut off (overflow)
  • Interactive charts ≥ static where applicable
  • Content parity (no missing slides/equations/text)
  • Visual quality ≥ Beamer
  • Slide centering stable; notation verbatim from Beamer

Workflow (single agent, two phases per round)

  1. Critic phase: Compare Beamer (.tex/.pdf) and Quarto (.qmd/.html) for the lecture. List every issue (Critical / Major / Minor). Save to quality_reports/[Lecture]_qa_critic_roundN.md. If no blocking issues, state "APPROVED" and stop.

  2. Fixer phase: Apply fixes from the critic report (Critical first, then Major, then Minor). Re-render Quarto. Do not self-approve.

  3. Re-audit: Run critic again on the updated Quarto. If not APPROVED and rounds < 5, repeat from step 2. After 5 rounds, report remaining issues to the user.

  4. Final report: Save quality_reports/[Lecture]_qa_final.md with gate status and any remaining issues.

Pre-flight

  • Locate Beamer and Quarto files; re-render if .qmd newer than .html; verify TikZ SVGs if used.
Install via CLI
npx skills add https://github.com/tesolchina/genAI2026 --skill qa-quarto
Repository Details
star Stars 7
call_split Forks 34
navigation Branch main
article Path SKILL.md
More from Creator