kwongfuk/codex-skills/paper-slides

Paper Slides: From Paper to Conference Talk

Generate conference presentation slides from: $ARGUMENTS

Context

This skill runs after Workflow 3 (/paper-writing). It takes a compiled paper and generates a presentation slide deck for conference oral talks, spotlight presentations, or poster lightning talks.

Unlike posters (single page, visual-first), slides tell a temporal story: each slide builds on the previous one, with progressive revelation of the research narrative. A good talk makes the audience understand why this matters before showing what was done.

Constants

• VENUE = `NeurIPS` — Target venue, determines color scheme. Supported: NeurIPS, ICML, ICLR, AAAI, ACL, EMNLP, CVPR, ECCV, GENERIC. Override via argument.
• TALK_TYPE = `spotlight` — Talk format. Options: oral (15-20 min), spotlight (5-8 min), poster-talk (3-5 min), invited (30-45 min). Determines slide count and content depth.
• TALK_MINUTES = 15 — Talk duration in minutes. Auto-adjusts slide count (~1 slide/minute for oral, ~1.5 slides/minute for spotlight). Override explicitly if needed.
• ASPECT_RATIO = `16:9` — Slide aspect ratio. Options: 16:9 (default, modern projectors), 4:3 (legacy).
• SPEAKER_NOTES = true — Generate \note{} blocks in beamer and corresponding PPTX notes. Set false for clean slides without notes.
• PAPER_DIR = `paper/` — Directory containing the compiled paper.
• OUTPUT_DIR = `slides/` — Output directory for all slide files.
• REVIEWER_MODEL = `gpt-5.4` — Model used via Codex MCP for slide review.
• AUTO_PROCEED = false — At each checkpoint, always wait for explicit user confirmation.
• COMPILER = `latexmk` — LaTeX build tool.
• ENGINE = `pdflatex` — LaTeX engine. Use xelatex for CJK text.

💡 Override: /paper-slides "paper/" — talk_type: oral, venue: ICML, minutes: 20, aspect: 4:3

Talk Type → Slide Count

Venue Color Schemes

Same as /paper-poster:

State Persistence (Compact Recovery)

Persist state to slides/SLIDES_STATE.json after each phase:

{
  "phase": 3,
  "venue": "NeurIPS",
  "talk_type": "spotlight",
  "slide_count": 10,
  "codex_thread_id": "019cfcf4-...",
  "status": "in_progress",
  "timestamp": "2026-03-18T15:00:00"
}

On startup: if SLIDES_STATE.json exists with "status": "in_progress" and within 24h → resume. Otherwise → fresh start.

Workflow

Phase 0: Input Validation & Setup

1. Check prerequisites:

``bash which pdflatex && which latexmk ``

1. Verify paper exists:

``bash ls $PAPER_DIR/main.tex || ls $PAPER_DIR/main.pdf ls $PAPER_DIR/sections/*.tex ls $PAPER_DIR/figures/ ``

1. Backup existing slides: if slides/ exists, copy to slides-backup-{timestamp}/

1. Create output directory: mkdir -p slides/figures

1. Detect CJK: if paper contains Chinese/Japanese/Korean, set ENGINE to xelatex

1. Determine slide count: from TALK_TYPE and TALK_MINUTES using the table above

1. Check for resume: read slides/SLIDES_STATE.json if it exists

State: Write SLIDES_STATE.json with phase: 0.

Phase 1: Content Extraction & Slide Outline

Read paper/sections/*.tex and build a slide-by-slide outline.

Slide template by talk type:

Oral (15-22 slides)

Spotlight (8-12 slides)

Poster-talk (5-8 slides)

For each slide, specify:

• Title (max 8 words)
• 3-5 bullet points (max 8 words each)
• Figure reference (if any) from paper/figures/
• Speaker note (2-3 sentences of what to say)
• Time allocation (in seconds)

Output: slides/SLIDE_OUTLINE.md

🚦 Checkpoint:

📊 Slide outline ready:
- Talk type: [TALK_TYPE] ([TALK_MINUTES] min)
- Slide count: [N] slides
- Figures used: [N] from paper/figures/
- Time budget: [breakdown]
 
Slide-by-slide outline:
1. [Title slide]
2. [Motivation — 1.5 min]
3. [Problem statement — 1 min]
...
 
Proceed to drafting? Or adjust the outline?

⛔ STOP HERE and wait for user response. This is the most critical checkpoint — the outline determines the entire talk flow.

Options:

• "go" → proceed to Phase 2
• adjustments (e.g., "merge slides 3-4", "add a demo slide", "cut the ablation") → revise
• "stop" → save to slides/SLIDE_OUTLINE.md

State: Write SLIDES_STATE.json with phase: 1.

Phase 2: Slide-by-Slide Content Drafting

For each slide in the outline, draft the actual content.

Presentation rules (enforced strictly):

For each slide, produce:

1. \frametitle{}
2. Content (itemize or figure + caption)
3. \note{} with speaker text (if SPEAKER_NOTES=true)

Phase 3: Generate Slides LaTeX

Create slides/main.tex using beamer.

Template structure:

\documentclass[aspectratio=169]{beamer}
 
% Venue theme
\usepackage{xcolor}
\definecolor{primary}{HTML}{VENUE_PRIMARY}
\definecolor{accent}{HTML}{VENUE_ACCENT}
 
% Clean theme
\usetheme{default}
\usecolortheme{default}
\setbeamercolor{frametitle}{fg=primary}
\setbeamercolor{title}{fg=primary}
\setbeamercolor{structure}{fg=accent}
\setbeamercolor{itemize item}{fg=primary}
\setbeamercolor{itemize subitem}{fg=accent}
\setbeamertemplate{navigation symbols}{}
\setbeamertemplate{footline}{
  \hfill\insertframenumber/\inserttotalframenumber\hspace{2mm}\vspace{2mm}
}
 
% Packages
\usepackage{graphicx,amsmath,booktabs}
\graphicspath{{figures/}}
 
% Speaker notes (if enabled)
% \setbeameroption{show notes on second screen=right}
 
% Metadata
\title{PAPER TITLE}
\author{Author 1 \and Author 2}
\institute{Affiliation}
\date{VENUE YEAR}
 
\begin{document}
 
\begin{frame}
\titlepage
\end{frame}
 
% Content slides follow...
 
\begin{frame}{Motivation}
\begin{itemize}
  \item Bullet point 1
  \item Bullet point 2
  \item \textbf{Key insight in bold}
\end{itemize}
\note{Speaker note: explain the motivation...}
\end{frame}
 
% Figure slide example
\begin{frame}{Method Overview}
\centering
\includegraphics[width=0.85\textwidth]{method_overview.pdf}
\vspace{0.5em}
\begin{itemize}
  \item Key annotation about the figure
\end{itemize}
\note{Walk through the figure left to right...}
\end{frame}
 
% ... more slides ...
 
\begin{frame}{Thank You}
\centering
{\Large Questions?}\\[2em]
Paper: [URL or QR placeholder]\\
Code: [URL or QR placeholder]
\end{frame}
 
\end{document}

Symlink figures:

ln -sf ../paper/figures/*.pdf slides/figures/ 2>/dev/null
ln -sf ../paper/figures/*.png slides/figures/ 2>/dev/null

Key formatting rules:

• Title font: ≥28pt, venue primary color
• Body font: ≥20pt
• Footnotes: ≥14pt
• No navigation symbols
• Frame numbers in bottom-right
• Clean white background (no gradients, no decorative elements)

Phase 4: Compile Slides

cd slides && latexmk -$ENGINE -interaction=nonstopmode main.tex

Error handling loop (max 3 attempts):

1. Parse error log
2. Fix: missing package, undefined command, file not found, overfull boxes
3. Recompile

Verification:

# Check slide count matches outline
pdfinfo slides/main.pdf | grep Pages

If page count differs significantly from outline (>2 slides off), investigate.

State: Write SLIDES_STATE.json with phase: 4.

Phase 5: Codex MCP Review

Send the slide outline + selected LaTeX frames to GPT-5.4 xhigh:

mcp__codex__codex:
  config: {"model_reasoning_effort": "xhigh"}
  prompt: |
    Review this [TALK_TYPE] presentation ([TALK_MINUTES] min) for [VENUE].
 
    Evaluate using these criteria (score 1-5 each):
 
    1. **Story arc** — Does the talk build a compelling narrative? (Problem → insight → method → evidence → takeaway)
    2. **Slide density** — Any slides with too much text? (Max 6 lines, 8 words/line)
    3. **Time budget** — Is [N] slides realistic for [TALK_MINUTES] minutes?
    4. **Figure visibility** — Will figures be readable on a projector?
    5. **Opening hook** — Do slides 2-3 grab attention? (Not "In this paper, we...")
    6. **Takeaway** — Is the final message clear and memorable?
    7. **Progressive build** — Are complex ideas revealed gradually?
 
    Slide outline:
    [PASTE SLIDE_OUTLINE.md]
 
    Selected frames (LaTeX):
    [PASTE KEY FRAMES]
 
    Provide:
    - Score for each criterion
    - Top 3 actionable fixes
    - Overall: Ready to present? (Yes / Needs revision / Major issues)

Apply fixes. Recompile if LaTeX was changed.

⚠️ If mcp__codex__codex is not available (no OpenAI API key), skip external review and proceed to Phase 6. Note the skip in SLIDES_STATE.json.

Save review to slides/SLIDES_REVIEW.md.

State: Write SLIDES_STATE.json with phase: 5.

Phase 6: Speaker Notes

For each slide, ensure a \note{} block exists with:

1. What to say (2-3 complete sentences, conversational tone)
2. Timing hint (e.g., "spend 1 minute here", "quick — 20 seconds")
3. Transition phrase to the next slide (e.g., "So how do we actually implement this? Let me show you...")

Also generate slides/speaker_notes.md as a standalone backup:

# Speaker Notes
 
## Slide 1: Title
[No speaking — wait for introduction]