[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7FD4STMPMNB3AE3YCRK4CVN":3,"skill-same:01M7FD4STMPMNB3AE3YCRK4CVN":48},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":32,"updatedAt":33,"install":34,"resources":47},"01M7FD4STMPMNB3AE3YCRK4CVN","paper-slides",null,"Generate conference presentation slides (beamer LaTeX → PDF + editable PPTX) from a compiled paper, with speaker notes and full talk script. Use when user says \"做PPT\", \"做幻灯片\", \"make slides\", \"conference talk\", \"presentation slides\", \"生成slides\", \"写演讲稿\", or wants beamer slides for a conference talk.","CZX-LBJ\u002Fdeepseek-science","CZX-LBJ","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F204594689?v=4","01M7FC59WQ9PFXT04MKGJDCFFT","czx","01M7FC59XB2GW0BKWN9JTDBQS7",1,"MIT","https:\u002F\u002Fgithub.com\u002FCZX-LBJ\u002Fdeepseek-science\u002Fblob\u002Fmain\u002Fpackages\u002Fskill\u002Fskill-science\u002Fassets\u002Fsources\u002Faris\u002Fskills\u002Fpaper-slides\u002FSKILL.md",[],[],"2026-11-16 10:05:34",false,"---\nname: paper-slides\ndescription: \"Generate conference presentation slides (beamer LaTeX → PDF + editable PPTX) from a compiled paper, with speaker notes and full talk script. Use when user says \\\"做PPT\\\", \\\"做幻灯片\\\", \\\"make slides\\\", \\\"conference talk\\\", \\\"presentation slides\\\", \\\"生成slides\\\", \\\"写演讲稿\\\", or wants beamer slides for a conference talk.\"\nargument-hint: \"[paper-directory-or-talk-length] [— style-ref: \u003Csource>]\"\nallowed-tools: Bash(*), Read, Write, Edit, Grep, Glob, mcp__codex__codex, mcp__codex__codex-reply\n---\n\n# Paper Slides: From Paper to Conference Talk\n\nGenerate conference presentation slides from: **$ARGUMENTS**\n\n## Context\n\nThis skill runs **after** Workflow 3 (`\u002Fpaper-writing`). It takes a compiled paper and generates a presentation slide deck for conference oral talks, spotlight presentations, or poster lightning talks.\n\nUnlike 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*.\n\n## Constants\n\n- **VENUE = `NeurIPS`** — Target venue, determines color scheme. Supported: `NeurIPS`, `ICML`, `ICLR`, `AAAI`, `ACL`, `EMNLP`, `CVPR`, `ECCV`, `GENERIC`. Override via argument.\n- **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.\n- **TALK_MINUTES = 15** — Talk duration in minutes. Auto-adjusts slide count (~1 slide\u002Fminute for oral, ~1.5 slides\u002Fminute for spotlight). Override explicitly if needed.\n- **ASPECT_RATIO = `16:9`** — Slide aspect ratio. Options: `16:9` (default, modern projectors), `4:3` (legacy).\n- **SPEAKER_NOTES = true** — Generate `\\note{}` blocks in beamer and corresponding PPTX notes. Set `false` for clean slides without notes.\n- **PAPER_DIR = `paper\u002F`** — Directory containing the compiled paper.\n- **OUTPUT_DIR = `slides\u002F`** — Output directory for all slide files.\n- **REVIEWER_MODEL = `gpt-6-astra`** — Model used via Codex MCP for slide review.\n- **AUTO_PROCEED = false** — At each checkpoint, **always wait for explicit user confirmation**.\n- **COMPILER = `latexmk`** — LaTeX build tool.\n- **ENGINE = `pdflatex`** — LaTeX engine. Use `xelatex` for CJK text.\n\n> 💡 Override: `\u002Fpaper-slides \"paper\u002F\" — talk_type: oral, venue: ICML, minutes: 20, aspect: 4:3`\n\n## Optional: Style reference (`— style-ref: \u003Csource>`, opt-in)\n\nLets the user steer the talk's **structural** rhythm (story beats, theorem density, figure density inherited from the source paper) toward a reference paper. **Default OFF — when the user does not pass `— style-ref`, do nothing differently from before.**\n\nOnly when `— style-ref: \u003Csource>` appears in `$ARGUMENTS`, run the helper FIRST:\n\n```bash\n# Resolve $STYLE_HELPER via the canonical strict-safe chain (see\n# shared-references\u002Fintegration-contract.md §2). Policy A — gate:\n# unresolved helper means --style-ref cannot be satisfied, so abort.\ncd \"$(git rev-parse --show-toplevel 2>\u002Fdev\u002Fnull || pwd)\" || exit 1\nif [ -z \"${ARIS_REPO:-}\" ] && [ -f .aris\u002Finstalled-skills.txt ]; then\n    ARIS_REPO=$(awk -F'\\t' '$1==\"repo_root\"{print $2; exit}' .aris\u002Finstalled-skills.txt 2>\u002Fdev\u002Fnull) || true\nfi\nif [ -z \"${ARIS_REPO:-}\" ] && [ -f \"$HOME\u002F.aris\u002Frepo\" ]; then\n    ARIS_REPO=$(cat \"$HOME\u002F.aris\u002Frepo\" 2>\u002Fdev\u002Fnull) || true\nfi\nSTYLE_HELPER=\".aris\u002Ftools\u002Fextract_paper_style.py\"\n[ -f \"$STYLE_HELPER\" ] || STYLE_HELPER=\"tools\u002Fextract_paper_style.py\"\n[ -f \"$STYLE_HELPER\" ] || { [ -n \"${ARIS_REPO:-}\" ] && STYLE_HELPER=\"$ARIS_REPO\u002Ftools\u002Fextract_paper_style.py\"; }\n[ -f \"$STYLE_HELPER\" ] || {\n  echo \"ERROR: extract_paper_style.py not resolved at .aris\u002Ftools\u002F, tools\u002F, \\$ARIS_REPO\u002Ftools\u002F, or via ~\u002F.aris\u002Frepo.\" >&2\n  echo \"       Fix: rerun bash tools\u002Finstall_aris.sh or smart_update.sh (refreshes ~\u002F.aris\u002Frepo), export ARIS_REPO, or copy the helper to tools\u002F.\" >&2\n  echo \"       --style-ref cannot be satisfied; aborting.\" >&2\n  exit 1\n}\nSTYLE_STATUS=0\nCACHE=$(python3 \"$STYLE_HELPER\" --source \"\u003Csource>\") || STYLE_STATUS=$?\ncase \"$STYLE_STATUS\" in\n  0) ;;                                       # use $CACHE\u002Fstyle_profile.md as structural guidance\n  2) echo \"warning: style-ref skipped (missing optional dep)\" >&2 ;;\n  3) echo \"error: --style-ref source failed; aborting slides\" >&2 ; exit 1 ;;\n  *) echo \"error: helper failed unexpectedly; aborting slides\" >&2 ; exit 1 ;;\nesac\n```\n\nSources accepted: local TeX dir \u002F file, local PDF, arXiv id, http(s) URL. Overleaf URLs\u002FIDs are rejected — clone via `\u002Foverleaf-sync setup \u003Cid>` first and pass the local clone path.\n\n**Strict rules** (full contract in `tools\u002Fextract_paper_style.py` docstring):\n\n- Use `style_profile.md` to align section-budget tendency and theorem-environment density. Talk-type slide count above still takes precedence.\n- **Never copy speaker-note prose, slide titles, or examples** from anything reachable through the cache. The talk content is from the user's paper, not the reference.\n- **Never pass `— style-ref` (or the cache contents) to the GPT-6-Astra reviewer sub-agent** — the reviewer must judge the talk's clarity on its own merits.\n\n## Talk Type → Slide Count\n\n| Talk Type | Duration | Slides | Content Depth |\n|-----------|----------|:------:|---------------|\n| `poster-talk` | 3-5 min | 5-8 | Problem + 1 method slide + 1 result + conclusion |\n| `spotlight` | 5-8 min | 8-12 | Problem + 2 method + 2 results + conclusion |\n| `oral` | 15-20 min | 15-22 | Full story with motivation, method detail, experiments, analysis |\n| `invited` | 30-45 min | 25-40 | Comprehensive: background, related work, deep method, extensive results, discussion |\n\n## Venue Color Schemes\n\nSame as `\u002Fpaper-poster-html`:\n\n| Venue | Primary | Accent | Background | Text |\n|-------|---------|--------|------------|------|\n| NeurIPS | `#8B5CF6` | `#2563EB` | `#FFFFFF` | `#1E1E1E` |\n| ICML | `#DC2626` | `#1D4ED8` | `#FFFFFF` | `#1E1E1E` |\n| ICLR | `#059669` | `#0284C7` | `#FFFFFF` | `#1E1E1E` |\n| CVPR | `#2563EB` | `#7C3AED` | `#FFFFFF` | `#1E1E1E` |\n| GENERIC | `#334155` | `#2563EB` | `#FFFFFF` | `#1E1E1E` |\n\n## State Persistence (Compact Recovery)\n\nPersist state to `slides\u002FSLIDES_STATE.json` after each phase:\n\n```json\n{\n  \"phase\": 3,\n  \"venue\": \"NeurIPS\",\n  \"talk_type\": \"spotlight\",\n  \"slide_count\": 10,\n  \"codex_thread_id\": \"019cfcf4-...\",\n  \"status\": \"in_progress\",\n  \"timestamp\": \"2026-03-18T15:00:00\"\n}\n```\n\n**On startup**: if `SLIDES_STATE.json` exists with `\"status\": \"in_progress\"` and within 24h → resume. Otherwise → fresh start.\n\n## Workflow\n\n### Phase 0: Input Validation & Setup\n\n1. **Check prerequisites**:\n   ```bash\n   which pdflatex && which latexmk\n   ```\n\n2. **Verify paper exists**:\n   ```bash\n   ls $PAPER_DIR\u002Fmain.tex || ls $PAPER_DIR\u002Fmain.pdf\n   ls $PAPER_DIR\u002Fsections\u002F*.tex\n   ls $PAPER_DIR\u002Ffigures\u002F\n   ```\n\n3. **Backup existing slides**: if `slides\u002F` exists, copy to `slides-backup-{timestamp}\u002F`\n\n4. **Create output directory**: `mkdir -p slides\u002Ffigures`\n\n5. **Detect CJK**: if paper contains Chinese\u002FJapanese\u002FKorean, set ENGINE to `xelatex`\n\n6. **Determine slide count**: from TALK_TYPE and TALK_MINUTES using the table above\n\n7. **Check for resume**: read `slides\u002FSLIDES_STATE.json` if it exists\n\n**State**: Write `SLIDES_STATE.json` with `phase: 0`.\n\n### Phase 1: Content Extraction & Slide Outline\n\nRead `paper\u002Fsections\u002F*.tex` and build a slide-by-slide outline.\n\n**Slide template by talk type**:\n\n#### Oral (15-22 slides)\n\n| Slide | Purpose | Content Source | Figure? |\n|:-----:|---------|----------------|:-------:|\n| 1 | Title | Paper metadata | No |\n| 2 | Outline | Section headers | No |\n| 3-4 | Motivation & Problem | Introduction | Optional |\n| 5 | Key Insight | Introduction (contribution) | No |\n| 6-9 | Method | Method section | Yes (hero figure) |\n| 10-14 | Results | Experiments | Yes (per slide) |\n| 15-16 | Analysis \u002F Ablations | Experiments | Yes |\n| 17 | Limitations | Conclusion | No |\n| 18 | Conclusion \u002F Takeaway | Conclusion | No |\n| 19 | Thank You + QR | — | QR code |\n\n#### Spotlight (8-12 slides)\n\n| Slide | Purpose | Content Source | Figure? |\n|:-----:|---------|----------------|:-------:|\n| 1 | Title | Paper metadata | No |\n| 2-3 | Problem + Why It Matters | Introduction | Optional |\n| 4 | Key Insight | Contribution | No |\n| 5-6 | Method | Method (condensed) | Yes (hero) |\n| 7-9 | Results | Key results only | Yes |\n| 10 | Takeaway | Conclusion | No |\n| 11 | Thank You + QR | — | QR code |\n\n#### Poster-talk (5-8 slides)\n\n| Slide | Purpose | Content Source | Figure? |\n|:-----:|---------|----------------|:-------:|\n| 1 | Title | Paper metadata | No |\n| 2 | Problem | Introduction (1 slide) | No |\n| 3 | Method | Method (1 slide) | Yes |\n| 4-5 | Results | Key result only | Yes |\n| 6 | Takeaway + QR | Conclusion | QR |\n\n**For each slide, specify**:\n- Title (max 8 words)\n- 3-5 bullet points (max 8 words each)\n- Figure reference (if any) from paper\u002Ffigures\u002F\n- Speaker note (2-3 sentences of what to say)\n- Time allocation (in seconds)\n\n**Output**: `slides\u002FSLIDE_OUTLINE.md`\n\n**🚦 Checkpoint:**\n\n```\n📊 Slide outline ready:\n- Talk type: [TALK_TYPE] ([TALK_MINUTES] min)\n- Slide count: [N] slides\n- Figures used: [N] from paper\u002Ffigures\u002F\n- Time budget: [breakdown]\n\nSlide-by-slide outline:\n1. [Title slide]\n2. [Motivation — 1.5 min]\n3. [Problem statement — 1 min]\n...\n\nProceed to drafting? Or adjust the outline?\n```\n\n**⛔ STOP HERE and wait for user response.** This is the most critical checkpoint — the outline determines the entire talk flow.\n\nOptions:\n- **\"go\"** → proceed to Phase 2\n- **adjustments** (e.g., \"merge slides 3-4\", \"add a demo slide\", \"cut the ablation\") → revise\n- **\"stop\"** → save to `slides\u002FSLIDE_OUTLINE.md`\n\n**State**: Write `SLIDES_STATE.json` with `phase: 1`.\n\n### Phase 2: Slide-by-Slide Content Drafting\n\nFor each slide in the outline, draft the actual content.\n\n**Presentation rules (enforced strictly)**:\n\n| Rule | Rationale |\n|------|-----------|\n| **One message per slide** | If a slide has two ideas, split it |\n| **Max 6 lines per slide** | More than 6 lines = wall of text |\n| **Max 8 words per line** | Audience reads, not listens, if text is long |\n| **Sentence fragments, not sentences** | \"Improves F1 by 3.2%\" not \"Our method improves the F1 score by 3.2 percentage points\" |\n| **Figure slides: figure ≥60% area** | The figure IS the content; bullets are annotations |\n| **Bold key numbers** | \"Achieves **94.3%** accuracy\" |\n| **Progressive disclosure** | Use `\\pause` or `\\onslide` for complex slides |\n| **No Related Work slide** | Unless invited talk (30+ min) |\n\n**For each slide, produce**:\n1. `\\frametitle{}`\n2. Content (itemize or figure + caption)\n3. `\\note{}` with speaker text (if SPEAKER_NOTES=true)\n\n### Phase 3: Generate Slides LaTeX\n\nCreate `slides\u002Fmain.tex` using beamer.\n\n**Template structure**:\n\n```latex\n\\documentclass[aspectratio=169]{beamer}\n\n% Venue theme\n\\usepackage{xcolor}\n\\definecolor{primary}{HTML}{VENUE_PRIMARY}\n\\definecolor{accent}{HTML}{VENUE_ACCENT}\n\n% Clean theme\n\\usetheme{default}\n\\usecolortheme{default}\n\\setbeamercolor{frametitle}{fg=primary}\n\\setbeamercolor{title}{fg=primary}\n\\setbeamercolor{structure}{fg=accent}\n\\setbeamercolor{itemize item}{fg=primary}\n\\setbeamercolor{itemize subitem}{fg=accent}\n\\setbeamertemplate{navigation symbols}{}\n\\setbeamertemplate{footline}{\n  \\hfill\\insertframenumber\u002F\\inserttotalframenumber\\hspace{2mm}\\vspace{2mm}\n}\n\n% Packages\n\\usepackage{graphicx,amsmath,booktabs}\n\\graphicspath{{figures\u002F}}\n\n% Speaker notes (if enabled)\n% \\setbeameroption{show notes on second screen=right}\n\n% Metadata\n\\title{PAPER TITLE}\n\\author{Author 1 \\and Author 2}\n\\institute{Affiliation}\n\\date{VENUE YEAR}\n\n\\begin{document}\n\n\\begin{frame}\n\\titlepage\n\\end{frame}\n\n% Content slides follow...\n\n\\begin{frame}{Motivation}\n\\begin{itemize}\n  \\item Bullet point 1\n  \\item Bullet point 2\n  \\item \\textbf{Key insight in bold}\n\\end{itemize}\n\\note{Speaker note: explain the motivation...}\n\\end{frame}\n\n% Figure slide example\n\\begin{frame}{Method Overview}\n\\centering\n\\includegraphics[width=0.85\\textwidth]{method_overview.pdf}\n\\vspace{0.5em}\n\\begin{itemize}\n  \\item Key annotation about the figure\n\\end{itemize}\n\\note{Walk through the figure left to right...}\n\\end{frame}\n\n% ... more slides ...\n\n\\begin{frame}{Thank You}\n\\centering\n{\\Large Questions?}\\\\[2em]\nPaper: [URL or QR placeholder]\\\\\nCode: [URL or QR placeholder]\n\\end{frame}\n\n\\end{document}\n```\n\n**Symlink figures**:\n```bash\nln -sf ..\u002Fpaper\u002Ffigures\u002F*.pdf slides\u002Ffigures\u002F 2>\u002Fdev\u002Fnull\nln -sf ..\u002Fpaper\u002Ffigures\u002F*.png slides\u002Ffigures\u002F 2>\u002Fdev\u002Fnull\n```\n\n**Key formatting rules**:\n- Title font: ≥28pt, venue primary color\n- Body font: ≥20pt\n- Footnotes: ≥14pt\n- No navigation symbols\n- Frame numbers in bottom-right\n- Clean white background (no gradients, no decorative elements)\n\n### Phase 4: Compile Slides\n\n```bash\ncd slides && latexmk -$ENGINE -interaction=nonstopmode main.tex\n```\n\n**Error handling loop** (max 3 attempts):\n1. Parse error log\n2. Fix: missing package, undefined command, file not found, overfull boxes\n3. Recompile\n\n**Verification**:\n```bash\n# Check slide count matches outline\npdfinfo slides\u002Fmain.pdf | grep Pages\n```\n\nIf page count differs significantly from outline (>2 slides off), investigate.\n\n**State**: Write `SLIDES_STATE.json` with `phase: 4`.\n\n### Phase 5: Codex MCP Review\n\nSend the slide outline + selected LaTeX frames to GPT-6-Astra xhigh:\n\n```\nmcp__codex__codex:\n  model: gpt-6-astra\n  config: {\"model_reasoning_effort\": \"xhigh\"}\n  prompt: |\n    Review this [TALK_TYPE] presentation ([TALK_MINUTES] min) for [VENUE].\n\n    Evaluate using these criteria (score 1-5 each):\n\n    1. **Story arc** — Does the talk build a compelling narrative? (Problem → insight → method → evidence → takeaway)\n    2. **Slide density** — Any slides with too much text? (Max 6 lines, 8 words\u002Fline)\n    3. **Time budget** — Is [N] slides realistic for [TALK_MINUTES] minutes?\n    4. **Figure visibility** — Will figures be readable on a projector?\n    5. **Opening hook** — Do slides 2-3 grab attention? (Not \"In this paper, we...\")\n    6. **Takeaway** — Is the final message clear and memorable?\n    7. **Progressive build** — Are complex ideas revealed gradually?\n\n    Slide outline:\n    [PASTE SLIDE_OUTLINE.md]\n\n    Selected frames (LaTeX):\n    [PASTE KEY FRAMES]\n\n    Provide:\n    - Score for each criterion\n    - Top 3 actionable fixes\n    - Overall: Ready to present? (Yes \u002F Needs revision \u002F Major issues)\n```\n\nApply fixes. Recompile if LaTeX was changed.\n\n> ⚠️ 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`.\n\nSave review to `slides\u002FSLIDES_REVIEW.md`.\n\n**State**: Write `SLIDES_STATE.json` with `phase: 5`.\n\n### Phase 6: Speaker Notes\n\nFor each slide, ensure a `\\note{}` block exists with:\n\n1. **What to say** (2-3 complete sentences, conversational tone)\n2. **Timing hint** (e.g., \"spend 1 minute here\", \"quick — 20 seconds\")\n3. **Transition phrase** to the next slide (e.g., \"So how do we actually implement this? Let me show you...\")\n\nAlso generate `slides\u002Fspeaker_notes.md` as a standalone backup:\n\n```markdown\n# Speaker Notes\n\n## Slide 1: Title\n[No speaking — wait for introduction]\n\n## Slide 2: Motivation\n\"Thank you. So let me start with the problem we're trying to solve...\"\n[Time: 1.5 min]\n\n## Slide 3: Problem Statement\n\"Specifically, the challenge is...\"\n→ Transition: \"To address this, our key insight is...\"\n[Time: 1 min]\n\n...\n```\n\n**State**: Write `SLIDES_STATE.json` with `phase: 6`.\n\n### Phase 7: PowerPoint Export\n\nGenerate an editable PPTX using `python-pptx`:\n\n```bash\npython3 -c \"import pptx\" 2>\u002Fdev\u002Fnull || pip install python-pptx\n```\n\nWrite `slides\u002Fgenerate_pptx.py` that:\n\n1. Creates a PPTX with correct aspect ratio (16:9 → 13.33\" x 7.5\"; 4:3 → 10\" x 7.5\")\n2. For each beamer frame:\n   - Creates a slide with matching layout\n   - Title in venue primary color, bold\n   - Bullet points with venue accent color markers\n   - Figures embedded as images (from slides\u002Ffigures\u002F)\n   - Speaker notes transferred to PPTX notes field\n3. Title slide with special formatting (centered, larger title)\n4. Thank You slide with centered text\n5. Applies venue color scheme throughout\n\n```bash\ncd slides && python3 generate_pptx.py\n# Output: slides\u002Fpresentation.pptx\n```\n\n> ⚠️ If `python-pptx` is not installed, skip with a note: \"Install `pip install python-pptx` to enable PowerPoint export.\"\n\n**State**: Write `SLIDES_STATE.json` with `phase: 7`.\n\n### Phase 8: Full Talk Script\n\nGenerate `slides\u002FTALK_SCRIPT.md` — a complete, word-for-word script for the talk.\n\nThis is different from speaker notes (brief reminders). The talk script is a **full manuscript** that can be read aloud or used for practice.\n\n```markdown\n# Talk Script: [Paper Title]\n\n**Venue**: [VENUE] [YEAR]\n**Talk type**: [TALK_TYPE] ([TALK_MINUTES] min)\n**Total slides**: [N]\n\n---\n\n## Slide 1: Title [0:00 - 0:15]\n\n*[Wait for chair introduction]*\n\n\"Thank you [chair name]. I'm [author] from [affiliation], and today I'll be talking about [short title].\"\n\n---\n\n## Slide 2: Motivation [0:15 - 1:30]\n\n\"Let me start with the problem. [Describe the real-world motivation in accessible terms]. This matters because [impact statement].\n\nThe current state of the art approaches this with [brief existing approach]. But there's a fundamental limitation: [gap statement].\"\n\n→ *Transition*: \"So what's our key insight?\"\n\n---\n\n## Slide 3: Key Insight [1:30 - 2:30]\n\n\"Our key observation is that [core insight in one sentence].\n\nThis leads us to propose [method name], which [one-sentence description].\"\n\n→ *Transition*: \"Let me walk you through how this works.\"\n\n---\n\n## Slide 4-N: [Continue for each slide...]\n\n...\n\n---\n\n## Slide [N]: Thank You [TALK_MINUTES:00]\n\n\"To summarize: we've shown that [main result]. The key takeaway is [memorable final message].\n\nThe paper and code are available at the QR code on screen. I'm happy to take questions.\"\n\n---\n\n## Time Budget Summary\n\n| Slide | Topic | Duration | Cumulative |\n|:-----:|-------|:--------:|:----------:|\n| 1 | Title | 0:15 | 0:15 |\n| 2 | Motivation | 1:15 | 1:30 |\n| 3 | Key Insight | 1:00 | 2:30 |\n| ... | ... | ... | ... |\n| N | Thank You | 0:15 | [TALK_MINUTES]:00 |\n\n**Total**: [sum] min (target: [TALK_MINUTES] min)\n\n---\n\n## Anticipated Q&A\n\n### Q1: How does this compare to [strongest baseline]?\n**A**: \"[Specific comparison with numbers]. Our advantage is particularly clear in [specific scenario], where we see [X%] improvement.\"\n\n### Q2: What are the main limitations?\n**A**: \"[Honest answer]. We see this as [future work direction].\"\n\n### Q3: How computationally expensive is this?\n**A**: \"[Training\u002Finference cost]. Compared to [baseline], our method requires [comparison].\"\n\n### Q4: Does this generalize to [related domain]?\n**A**: \"[Answer based on paper's discussion section].\"\n\n### Q5: What's the most surprising finding?\n**A**: \"[Interesting insight from the experiments].\"\n\n### Q6: How sensitive is the method to [hyperparameter\u002Fdesign choice]?\n**A**: \"[Reference ablation study if available].\"\n\n### Q7: What's the next step for this research?\n**A**: \"[Future work from conclusion].\"\n\n### Q8: [Domain-specific question]\n**A**: \"[Answer].\"\n```\n\n### Final Output Summary\n\n```\n📊 Slide generation complete:\n- Talk type: [TALK_TYPE] ([TALK_MINUTES] min) for [VENUE]\n- Files:\n  slides\u002F\n  ├── main.tex              # Beamer LaTeX source\n  ├── main.pdf              # Compiled slides (primary output)\n  ├── presentation.pptx     # Editable PowerPoint\n  ├── SLIDE_OUTLINE.md      # Slide-by-slide outline\n  ├── SLIDES_REVIEW.md      # GPT-6-Astra review feedback\n  ├── speaker_notes.md      # Per-slide speaker notes\n  ├── TALK_SCRIPT.md        # Full word-for-word talk script + Q&A\n  ├── SLIDES_STATE.json     # State persistence\n  ├── generate_pptx.py      # PPTX generation script\n  └── figures\u002F              # Symlinked from paper\u002Ffigures\u002F\n\nNext steps:\n1. Practice with TALK_SCRIPT.md (read aloud, time yourself)\n2. Edit presentation.pptx for visual tweaks (animations, custom graphics)\n3. Review Anticipated Q&A section before the talk\n4. Do a dry run with a colleague\n```\n\n**State**: Write `SLIDES_STATE.json` with `phase: 8, status: \"completed\"`.\n\n## Recommended Follow-up: `\u002Fslides-polish`\n\nAfter this skill produces the initial Beamer + PPTX, the typical drift is\n**typography proportion + per-slide layout**, not content. Run\n`\u002Fslides-polish` as a focused post-generation polish phase: it does\nper-page Codex review against a reference visual (e.g., a prior academic\ntalk), bumps PPTX fonts to projector-readable sizes, fixes text-frame\noverflow, and applies a fix-pattern catalog (italic style leaks, em-dash\nspacing, image aspect ratio, Chinese-font hints, anonymity placeholders).\n\n`\u002Fslides-polish` is read-only on content (no claim \u002F number \u002F citation\nedits) and preserves speaker notes verbatim. Invocation:\n\n```\n\u002Fslides-polish slides\u002F — reference: \u003Cref-pdf> [— style: generic | why-rf | \u003Cvenue>]\n```\n\nSkip it for short decks (\u003C 5 slides) or when a complete redesign is\nneeded (re-run `\u002Fpaper-slides` instead).\n\n## Key Rules\n\n- **Large file handling**: If the Write tool fails due to file size, immediately retry using Bash (`cat \u003C\u003C 'EOF' > file`) to write in chunks. Do NOT ask the user for permission — just do it silently.\n- **One message per slide.** If a slide has two ideas, split it into two slides.\n- **Do NOT fabricate data.** All numbers must come from `paper\u002Fsections\u002F*.tex`.\n- **Bullet points only** — never full sentences on slides. Sentence fragments are fine.\n- **Figure slides: figure ≥60% of slide area.** The figure IS the content.\n- **Progressive disclosure**: Use `\\pause` or `\\onslide` for complex method slides.\n- **De-AI polish**: Remove watch words from all slide text and talk script.\n- **Do NOT hallucinate citations.** Reference only papers cited in the paper.\n- **Opening hook matters**: Never start with \"In this paper, we...\" — start with the problem or a provocative question.\n- **Font size minimums**: Title ≥28pt, body ≥20pt, footnotes ≥14pt.\n- **Feishu notifications are optional.** If `~\u002F.claude\u002Ffeishu.json` exists, send notifications. If absent, skip.\n\n## Parameter Pass-Through\n\n```\n\u002Fpaper-slides \"paper\u002F\" — talk_type: oral, venue: ICML, minutes: 20, aspect: 4:3, notes: false\n```\n\n| Parameter | Default | Description |\n|-----------|---------|-------------|\n| `venue` | NeurIPS | Conference for color scheme |\n| `talk_type` | spotlight | oral\u002Fspotlight\u002Fposter-talk\u002Finvited |\n| `minutes` | 15 | Talk duration |\n| `aspect` | 16:9 | Aspect ratio (16:9 \u002F 4:3) |\n| `notes` | true | Generate speaker notes |\n| `engine` | pdflatex | LaTeX engine |\n| `auto proceed` | false | Skip checkpoints |\n","# Paper Slides: From Paper to Conference Talk\n\nGenerate conference presentation slides from: **$ARGUMENTS**\n\n## Context\n\nThis skill runs **after** Workflow 3 (`\u002Fpaper-writing`). It takes a compiled paper and generates a presentation slide deck for conference oral talks, spotlight presentations, or poster lightning talks.\n\nUnlike 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*.\n\n## Constants\n\n- **VENUE = `NeurIPS`** — Target venue, determines color scheme. Supported: `NeurIPS`, `ICML`, `ICLR`, `AAAI`, `ACL`, `EMNLP`, `CVPR`, `ECCV`, `GENERIC`. Override via argument.\n- **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.\n- **TALK_MINUTES = 15** — Talk duration in minutes. Auto-adjusts slide count (~1 slide\u002Fminute for oral, ~1.5 slides\u002Fminute for spotlight). Override explicitly if needed.\n- **ASPECT_RATIO = `16:9`** — Slide aspect ratio. Options: `16:9` (default, modern projectors), `4:3` (legacy).\n- **SPEAKER_NOTES = true** — Generate `\\note{}` blocks in beamer and corresponding PPTX notes. Set `false` for clean slides without notes.\n- **PAPER_DIR = `paper\u002F`** — Directory containing the compiled paper.\n- **OUTPUT_DIR = `slides\u002F`** — Output directory for all slide files.\n- **REVIEWER_MODEL = `gpt-6-astra`** — Model used via Codex MCP for slide review.\n- **AUTO_PROCEED = false** — At each checkpoint, **always wait for explicit user confirmation**.\n- **COMPILER = `latexmk`** — LaTeX build tool.\n- **ENGINE = `pdflatex`** — LaTeX engine. Use `xelatex` for CJK text.\n\n> 💡 Override: `\u002Fpaper-slides \"paper\u002F\" — talk_type: oral, venue: ICML, minutes: 20, aspect: 4:3`\n\n## Optional: Style reference (`— style-ref: \u003Csource>`, opt-in)\n\nLets the user steer the talk's **structural** rhythm (story beats, theorem density, figure density inherited from the source paper) toward a reference paper. **Default OFF — when the user does not pass `— style-ref`, do nothing differently from before.**\n\nOnly when `— style-ref: \u003Csource>` appears in `$ARGUMENTS`, run the helper FIRST:\n\n```bash\n# Resolve $STYLE_HELPER via the canonical strict-safe chain (see\n# shared-references\u002Fintegration-contract.md §2). Policy A — gate:\n# unresolved helper means --style-ref cannot be satisfied, so abort.\ncd \"$(git rev-parse --show-toplevel 2>\u002Fdev\u002Fnull || pwd)\" || exit 1\nif [ -z \"${ARIS_REPO:-}\" ] && [ -f .aris\u002Finstalled-skills.txt ]; then\n    ARIS_REPO=$(awk -F'\\t' '$1==\"repo_root\"{print $2; exit}' .aris\u002Finstalled-skills.txt 2>\u002Fdev\u002Fnull) || true\nfi\nif [ -z \"${ARIS_REPO:-}\" ] && [ -f \"$HOME\u002F.aris\u002Frepo\" ]; then\n    ARIS_REPO=$(cat \"$HOME\u002F.aris\u002Frepo\" 2>\u002Fdev\u002Fnull) || true\nfi\nSTYLE_HELPER=\".aris\u002Ftools\u002Fextract_paper_style.py\"\n[ -f \"$STYLE_HELPER\" ] || STYLE_HELPER=\"tools\u002Fextract_paper_style.py\"\n[ -f \"$STYLE_HELPER\" ] || { [ -n \"${ARIS_REPO:-}\" ] && STYLE_HELPER=\"$ARIS_REPO\u002Ftools\u002Fextract_paper_style.py\"; }\n[ -f \"$STYLE_HELPER\" ] || {\n  echo \"ERROR: extract_paper_style.py not resolved at .aris\u002Ftools\u002F, tools\u002F, \\$ARIS_REPO\u002Ftools\u002F, or via ~\u002F.aris\u002Frepo.\" >&2\n  echo \"       Fix: rerun bash tools\u002Finstall_aris.sh or smart_update.sh (refreshes ~\u002F.aris\u002Frepo), export ARIS_REPO, or copy the helper to tools\u002F.\" >&2\n  echo \"       --style-ref cannot be satisfied; aborting.\" >&2\n  exit 1\n}\nSTYLE_STATUS=0\nCACHE=$(python3 \"$STYLE_HELPER\" --source \"\u003Csource>\") || STYLE_STATUS=$?\ncase \"$STYLE_STATUS\" in\n  0) ;;                                       # use $CACHE\u002Fstyle_profile.md as structural guidance\n  2) echo \"warning: style-ref skipped (missing optional dep)\" >&2 ;;\n  3) echo \"error: --style-ref source failed; aborting slides\" >&2 ; exit 1 ;;\n  *) echo \"error: helper failed unexpectedly; aborting slides\" >&2 ; exit 1 ;;\nesac\n```\n\nSources accepted: local TeX dir \u002F file, local PDF, arXiv id, http(s) URL. Overleaf URLs\u002FIDs are rejected — clone via `\u002Foverleaf-sync setup \u003Cid>` first and pass the local clone path.\n\n**Strict rules** (full contract in `tools\u002Fextract_paper_style.py` docstring):\n\n- Use `style_profile.md` to align section-budget tendency and theorem-environment density. Talk-type slide count above still takes precedence.\n- **Never copy speaker-note prose, slide titles, or examples** from anything reachable through the cache. The talk content is from the user's paper, not the reference.\n- **Never pass `— style-ref` (or the cache contents) to the GPT-6-Astra reviewer sub-agent** — the reviewer must judge the talk's clarity on its own merits.\n\n## Talk Type → Slide Count\n\n| Talk Type | Duration | Slides | Content Depth |\n|-----------|----------|:------:|---------------|\n| `poster-talk` | 3-5 min | 5-8 | Problem + 1 method slide + 1 result + conclusion |\n| `spotlight` | 5-8 min | 8-12 | Problem + 2 method + 2 results + conclusion |\n| `oral` | 15-20 min | 15-22 | Full story with motivation, method detail, experiments, analysis |\n| `invited` | 30-45 min | 25-40 | Comprehensive: background, related work, deep method, extensive results, discussion |\n\n## Venue Color Schemes\n\nSame as `\u002Fpaper-poster-html`:\n\n| Venue | Primary | Accent | Background | Text |\n|-------|---------|--------|------------|------|\n| NeurIPS | `#8B5CF6` | `#2563EB` | `#FFFFFF` | `#1E1E1E` |\n| ICML | `#DC2626` | `#1D4ED8` | `#FFFFFF` | `#1E1E1E` |\n| ICLR | `#059669` | `#0284C7` | `#FFFFFF` | `#1E1E1E` |\n| CVPR | `#2563EB` | `#7C3AED` | `#FFFFFF` | `#1E1E1E` |\n| GENERIC | `#334155` | `#2563EB` | `#FFFFFF` | `#1E1E1E` |\n\n## State Persistence (Compact Recovery)\n\nPersist state to `slides\u002FSLIDES_STATE.json` after each phase:\n\n```json\n{\n  \"phase\": 3,\n  \"venue\": \"NeurIPS\",\n  \"talk_type\": \"spotlight\",\n  \"slide_count\": 10,\n  \"codex_thread_id\": \"019cfcf4-...\",\n  \"status\": \"in_progress\",\n  \"timestamp\": \"2026-03-18T15:00:00\"\n}\n```\n\n**On startup**: if `SLIDES_STATE.json` exists with `\"status\": \"in_progress\"` and within 24h → resume. Otherwise → fresh start.\n\n## Workflow\n\n### Phase 0: Input Validation & Setup\n\n1. **Check prerequisites**:\n   ```bash\n   which pdflatex && which latexmk\n   ```\n\n2. **Verify paper exists**:\n   ```bash\n   ls $PAPER_DIR\u002Fmain.tex || ls $PAPER_DIR\u002Fmain.pdf\n   ls $PAPER_DIR\u002Fsections\u002F*.tex\n   ls $PAPER_DIR\u002Ffigures\u002F\n   ```\n\n3. **Backup existing slides**: if `slides\u002F` exists, copy to `slides-backup-{timestamp}\u002F`\n\n4. **Create output directory**: `mkdir -p slides\u002Ffigures`\n\n5. **Detect CJK**: if paper contains Chinese\u002FJapanese\u002FKorean, set ENGINE to `xelatex`\n\n6. **Determine slide count**: from TALK_TYPE and TALK_MINUTES using the table above\n\n7. **Check for resume**: read `slides\u002FSLIDES_STATE.json` if it exists\n\n**State**: Write `SLIDES_STATE.json` with `phase: 0`.\n\n### Phase 1: Content Extraction & Slide Outline\n\nRead `paper\u002Fsections\u002F*.tex` and build a slide-by-slide outline.\n\n**Slide template by talk type**:\n\n#### Oral (15-22 slides)\n\n| Slide | Purpose | Content Source | Figure? |\n|:-----:|---------|----------------|:-------:|\n| 1 | Title | Paper metadata | No |\n| 2 | Outline | Section headers | No |\n| 3-4 | Motivation & Problem | Introduction | Optional |\n| 5 | Key Insight | Introduction (contribution) | No |\n| 6-9 | Method | Method section | Yes (hero figure) |\n| 10-14 | Results | Experiments | Yes (per slide) |\n| 15-16 | Analysis \u002F Ablations | Experiments | Yes |\n| 17 | Limitations | Conclusion | No |\n| 18 | Conclusion \u002F Takeaway | Conclusion | No |\n| 19 | Thank You + QR | — | QR code |\n\n#### Spotlight (8-12 slides)\n\n| Slide | Purpose | Content Source | Figure? |\n|:-----:|---------|----------------|:-------:|\n| 1 | Title | Paper metadata | No |\n| 2-3 | Problem + Why It Matters | Introduction | Optional |\n| 4 | Key Insight | Contribution | No |\n| 5-6 | Method | Method (condensed) | Yes (hero) |\n| 7-9 | Results | Key results only | Yes |\n| 10 | Takeaway | Conclusion | No |\n| 11 | Thank You + QR | — | QR code |\n\n#### Poster-talk (5-8 slides)\n\n| Slide | Purpose | Content Source | Figure? |\n|:-----:|---------|----------------|:-------:|\n| 1 | Title | Paper metadata | No |\n| 2 | Problem | Introduction (1 slide) | No |\n| 3 | Method | Method (1 slide) | Yes |\n| 4-5 | Results | Key result only | Yes |\n| 6 | Takeaway + QR | Conclusion | QR |\n\n**For each slide, specify**:\n- Title (max 8 words)\n- 3-5 bullet points (max 8 words each)\n- Figure reference (if any) from paper\u002Ffigures\u002F\n- Speaker note (2-3 sentences of what to say)\n- Time allocation (in seconds)\n\n**Output**: `slides\u002FSLIDE_OUTLINE.md`\n\n**🚦 Checkpoint:**\n\n```\n📊 Slide outline ready:\n- Talk type: [TALK_TYPE] ([TALK_MINUTES] min)\n- Slide count: [N] slides\n- Figures used: [N] from paper\u002Ffigures\u002F\n- Time budget: [breakdown]\n\nSlide-by-slide outline:\n1. [Title slide]\n2. [Motivation — 1.5 min]\n3. [Problem statement — 1 min]\n...\n\nProceed to drafting? Or adjust the outline?\n```\n\n**⛔ STOP HERE and wait for user response.** This is the most critical checkpoint — the outline determines the entire talk flow.\n\nOptions:\n- **\"go\"** → proceed to Phase 2\n- **adjustments** (e.g., \"merge slides 3-4\", \"add a demo slide\", \"cut the ablation\") → revise\n- **\"stop\"** → save to `slides\u002FSLIDE_OUTLINE.md`\n\n**State**: Write `SLIDES_STATE.json` with `phase: 1`.\n\n### Phase 2: Slide-by-Slide Content Drafting\n\nFor each slide in the outline, draft the actual content.\n\n**Presentation rules (enforced strictly)**:\n\n| Rule | Rationale |\n|------|-----------|\n| **One message per slide** | If a slide has two ideas, split it |\n| **Max 6 lines per slide** | More than 6 lines = wall of text |\n| **Max 8 words per line** | Audience reads, not listens, if text is long |\n| **Sentence fragments, not sentences** | \"Improves F1 by 3.2%\" not \"Our method improves the F1 score by 3.2 percentage points\" |\n| **Figure slides: figure ≥60% area** | The figure IS the content; bullets are annotations |\n| **Bold key numbers** | \"Achieves **94.3%** accuracy\" |\n| **Progressive disclosure** | Use `\\pause` or `\\onslide` for complex slides |\n| **No Related Work slide** | Unless invited talk (30+ min) |\n\n**For each slide, produce**:\n1. `\\frametitle{}`\n2. Content (itemize or figure + caption)\n3. `\\note{}` with speaker text (if SPEAKER_NOTES=true)\n\n### Phase 3: Generate Slides LaTeX\n\nCreate `slides\u002Fmain.tex` using beamer.\n\n**Template structure**:\n\n```latex\n\\documentclass[aspectratio=169]{beamer}\n\n% Venue theme\n\\usepackage{xcolor}\n\\definecolor{primary}{HTML}{VENUE_PRIMARY}\n\\definecolor{accent}{HTML}{VENUE_ACCENT}\n\n% Clean theme\n\\usetheme{default}\n\\usecolortheme{default}\n\\setbeamercolor{frametitle}{fg=primary}\n\\setbeamercolor{title}{fg=primary}\n\\setbeamercolor{structure}{fg=accent}\n\\setbeamercolor{itemize item}{fg=primary}\n\\setbeamercolor{itemize subitem}{fg=accent}\n\\setbeamertemplate{navigation symbols}{}\n\\setbeamertemplate{footline}{\n  \\hfill\\insertframenumber\u002F\\inserttotalframenumber\\hspace{2mm}\\vspace{2mm}\n}\n\n% Packages\n\\usepackage{graphicx,amsmath,booktabs}\n\\graphicspath{{figures\u002F}}\n\n% Speaker notes (if enabled)\n% \\setbeameroption{show notes on second screen=right}\n\n% Metadata\n\\title{PAPER TITLE}\n\\author{Author 1 \\and Author 2}\n\\institute{Affiliation}\n\\date{VENUE YEAR}\n\n\\begin{document}\n\n\\begin{frame}\n\\titlepage\n\\end{frame}\n\n% Content slides follow...\n\n\\begin{frame}{Motivation}\n\\begin{itemize}\n  \\item Bullet point 1\n  \\item Bullet point 2\n  \\item \\textbf{Key insight in bold}\n\\end{itemize}\n\\note{Speaker note: explain the motivation...}\n\\end{frame}\n\n% Figure slide example\n\\begin{frame}{Method Overview}\n\\centering\n\\includegraphics[width=0.85\\textwidth]{method_overview.pdf}\n\\vspace{0.5em}\n\\begin{itemize}\n  \\item Key annotation about the figure\n\\end{itemize}\n\\note{Walk through the figure left to right...}\n\\end{frame}\n\n% ... more slides ...\n\n\\begin{frame}{Thank You}\n\\centering\n{\\Large Questions?}\\\\[2em]\nPaper: [URL or QR placeholder]\\\\\nCode: [URL or QR placeholder]\n\\end{frame}\n\n\\end{document}\n```\n\n**Symlink figures**:\n```bash\nln -sf ..\u002Fpaper\u002Ffigures\u002F*.pdf slides\u002Ffigures\u002F 2>\u002Fdev\u002Fnull\nln -sf ..\u002Fpaper\u002Ffigures\u002F*.png slides\u002Ffigures\u002F 2>\u002Fdev\u002Fnull\n```\n\n**Key formatting rules**:\n- Title font: ≥28pt, venue primary color\n- Body font: ≥20pt\n- Footnotes: ≥14pt\n- No navigation symbols\n- Frame numbers in bottom-right\n- Clean white background (no gradients, no decorative elements)\n\n### Phase 4: Compile Slides\n\n```bash\ncd slides && latexmk -$ENGINE -interaction=nonstopmode main.tex\n```\n\n**Error handling loop** (max 3 attempts):\n1. Parse error log\n2. Fix: missing package, undefined command, file not found, overfull boxes\n3. Recompile\n\n**Verification**:\n```bash\n# Check slide count matches outline\npdfinfo slides\u002Fmain.pdf | grep Pages\n```\n\nIf page count differs significantly from outline (>2 slides off), investigate.\n\n**State**: Write `SLIDES_STATE.json` with `phase: 4`.\n\n### Phase 5: Codex MCP Review\n\nSend the slide outline + selected LaTeX frames to GPT-6-Astra xhigh:\n\n```\nmcp__codex__codex:\n  model: gpt-6-astra\n  config: {\"model_reasoning_effort\": \"xhigh\"}\n  prompt: |\n    Review this [TALK_TYPE] presentation ([TALK_MINUTES] min) for [VENUE].\n\n    Evaluate using these criteria (score 1-5 each):\n\n    1. **Story arc** — Does the talk build a compelling narrative? (Problem → insight → method → evidence → takeaway)\n    2. **Slide density** — Any slides with too much text? (Max 6 lines, 8 words\u002Fline)\n    3. **Time budget** — Is [N] slides realistic for [TALK_MINUTES] minutes?\n    4. **Figure visibility** — Will figures be readable on a projector?\n    5. **Opening hook** — Do slides 2-3 grab attention? (Not \"In this paper, we...\")\n    6. **Takeaway** — Is the final message clear and memorable?\n    7. **Progressive build** — Are complex ideas revealed gradually?\n\n    Slide outline:\n    [PASTE SLIDE_OUTLINE.md]\n\n    Selected frames (LaTeX):\n    [PASTE KEY FRAMES]\n\n    Provide:\n    - Score for each criterion\n    - Top 3 actionable fixes\n    - Overall: Ready to present? (Yes \u002F Needs revision \u002F Major issues)\n```\n\nApply fixes. Recompile if LaTeX was changed.\n\n> ⚠️ 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`.\n\nSave review to `slides\u002FSLIDES_REVIEW.md`.\n\n**State**: Write `SLIDES_STATE.json` with `phase: 5`.\n\n### Phase 6: Speaker Notes\n\nFor each slide, ensure a `\\note{}` block exists with:\n\n1. **What to say** (2-3 complete sentences, conversational tone)\n2. **Timing hint** (e.g., \"spend 1 minute here\", \"quick — 20 seconds\")\n3. **Transition phrase** to the next slide (e.g., \"So how do we actually implement this? Let me show you...\")\n\nAlso generate `slides\u002Fspeaker_notes.md` as a standalone backup:\n\n```markdown\n# Speaker Notes\n\n## Slide 1: Title\n[No speaking — wait for introduction]\n\n## Slide 2: Motivation\n\"Thank you. So let me start with the problem we're trying to solve...\"\n[Time: 1.5 min]\n\n## Slide 3: Problem Statement\n\"Specifically, the challenge is...\"\n→ Transition: \"To address this, our key insight is...\"\n[Time: 1 min]\n\n...\n```\n\n**State**: Write `SLIDES_STATE.json` with `phase: 6`.\n\n### Phase 7: PowerPoint Export\n\nGenerate an editable PPTX using `python-pptx`:\n\n```bash\npython3 -c \"import pptx\" 2>\u002Fdev\u002Fnull || pip install python-pptx\n```\n\nWrite `slides\u002Fgenerate_pptx.py` that:\n\n1. Creates a PPTX with correct aspect ratio (16:9 → 13.33\" x 7.5\"; 4:3 → 10\" x 7.5\")\n2. For each beamer frame:\n   - Creates a slide with matching layout\n   - Title in venue primary color, bold\n   - Bullet points with venue accent color markers\n   - Figures embedded as images (from slides\u002Ffigures\u002F)\n   - Speaker notes transferred to PPTX notes field\n3. Title slide with special formatting (centered, larger title)\n4. Thank You slide with centered text\n5. Applies venue color scheme throughout\n\n```bash\ncd slides && python3 generate_pptx.py\n# Output: slides\u002Fpresentation.pptx\n```\n\n> ⚠️ If `python-pptx` is not installed, skip with a note: \"Install `pip install python-pptx` to enable PowerPoint export.\"\n\n**State**: Write `SLIDES_STATE.json` with `phase: 7`.\n\n### Phase 8: Full Talk Script\n\nGenerate `slides\u002FTALK_SCRIPT.md` — a complete, word-for-word script for the talk.\n\nThis is different from speaker notes (brief reminders). The talk script is a **full manuscript** that can be read aloud or used for practice.\n\n```markdown\n# Talk Script: [Paper Title]\n\n**Venue**: [VENUE] [YEAR]\n**Talk type**: [TALK_TYPE] ([TALK_MINUTES] min)\n**Total slides**: [N]\n\n---\n\n## Slide 1: Title [0:00 - 0:15]\n\n*[Wait for chair introduction]*\n\n\"Thank you [chair name]. I'm [author] from [affiliation], and today I'll be talking about [short title].\"\n\n---\n\n## Slide 2: Motivation [0:15 - 1:30]\n\n\"Let me start with the problem. [Describe the real-world motivation in accessible terms]. This matters because [impact statement].\n\nThe current state of the art approaches this with [brief existing approach]. But there's a fundamental limitation: [gap statement].\"\n\n→ *Transition*: \"So what's our key insight?\"\n\n---\n\n## Slide 3: Key Insight [1:30 - 2:30]\n\n\"Our key observation is that [core insight in one sentence].\n\nThis leads us to propose [method name], which [one-sentence description].\"\n\n→ *Transition*: \"Let me walk you through how this works.\"\n\n---\n\n## Slide 4-N: [Continue for each slide...]\n\n...\n\n---\n\n## Slide [N]: Thank You [TALK_MINUTES:00]\n\n\"To summarize: we've shown that [main result]. The key takeaway is [memorable final message].\n\nThe paper and code are available at the QR code on screen. I'm happy to take questions.\"\n\n---\n\n## Time Budget Summary\n\n| Slide | Topic | Duration | Cumulative |\n|:-----:|-------|:--------:|:----------:|\n| 1 | Title | 0:15 | 0:15 |\n| 2 | Motivation | 1:15 | 1:30 |\n| 3 | Key Insight | 1:00 | 2:30 |\n| ... | ... | ... | ... |\n| N | Thank You | 0:15 | [TALK_MINUTES]:00 |\n\n**Total**: [sum] min (target: [TALK_MINUTES] min)\n\n---\n\n## Anticipated Q&A\n\n### Q1: How does this compare to [strongest baseline]?\n**A**: \"[Specific comparison with numbers]. Our advantage is particularly clear in [specific scenario], where we see [X%] improvement.\"\n\n### Q2: What are the main limitations?\n**A**: \"[Honest answer]. We see this as [future work direction].\"\n\n### Q3: How computationally expensive is this?\n**A**: \"[Training\u002Finference cost]. Compared to [baseline], our method requires [comparison].\"\n\n### Q4: Does this generalize to [related domain]?\n**A**: \"[Answer based on paper's discussion section].\"\n\n### Q5: What's the most surprising finding?\n**A**: \"[Interesting insight from the experiments].\"\n\n### Q6: How sensitive is the method to [hyperparameter\u002Fdesign choice]?\n**A**: \"[Reference ablation study if available].\"\n\n### Q7: What's the next step for this research?\n**A**: \"[Future work from conclusion].\"\n\n### Q8: [Domain-specific question]\n**A**: \"[Answer].\"\n```\n\n### Final Output Summary\n\n```\n📊 Slide generation complete:\n- Talk type: [TALK_TYPE] ([TALK_MINUTES] min) for [VENUE]\n- Files:\n  slides\u002F\n  ├── main.tex              # Beamer LaTeX source\n  ├── main.pdf              # Compiled slides (primary output)\n  ├── presentation.pptx     # Editable PowerPoint\n  ├── SLIDE_OUTLINE.md      # Slide-by-slide outline\n  ├── SLIDES_REVIEW.md      # GPT-6-Astra review feedback\n  ├── speaker_notes.md      # Per-slide speaker notes\n  ├── TALK_SCRIPT.md        # Full word-for-word talk script + Q&A\n  ├── SLIDES_STATE.json     # State persistence\n  ├── generate_pptx.py      # PPTX generation script\n  └── figures\u002F              # Symlinked from paper\u002Ffigures\u002F\n\nNext steps:\n1. Practice with TALK_SCRIPT.md (read aloud, time yourself)\n2. Edit presentation.pptx for visual tweaks (animations, custom graphics)\n3. Review Anticipated Q&A section before the talk\n4. Do a dry run with a colleague\n```\n\n**State**: Write `SLIDES_STATE.json` with `phase: 8, status: \"completed\"`.\n\n## Recommended Follow-up: `\u002Fslides-polish`\n\nAfter this skill produces the initial Beamer + PPTX, the typical drift is\n**typography proportion + per-slide layout**, not content. Run\n`\u002Fslides-polish` as a focused post-generation polish phase: it does\nper-page Codex review against a reference visual (e.g., a prior academic\ntalk), bumps PPTX fonts to projector-readable sizes, fixes text-frame\noverflow, and applies a fix-pattern catalog (italic style leaks, em-dash\nspacing, image aspect ratio, Chinese-font hints, anonymity placeholders).\n\n`\u002Fslides-polish` is read-only on content (no claim \u002F number \u002F citation\nedits) and preserves speaker notes verbatim. Invocation:\n\n```\n\u002Fslides-polish slides\u002F — reference: \u003Cref-pdf> [— style: generic | why-rf | \u003Cvenue>]\n```\n\nSkip it for short decks (\u003C 5 slides) or when a complete redesign is\nneeded (re-run `\u002Fpaper-slides` instead).\n\n## Key Rules\n\n- **Large file handling**: If the Write tool fails due to file size, immediately retry using Bash (`cat \u003C\u003C 'EOF' > file`) to write in chunks. Do NOT ask the user for permission — just do it silently.\n- **One message per slide.** If a slide has two ideas, split it into two slides.\n- **Do NOT fabricate data.** All numbers must come from `paper\u002Fsections\u002F*.tex`.\n- **Bullet points only** — never full sentences on slides. Sentence fragments are fine.\n- **Figure slides: figure ≥60% of slide area.** The figure IS the content.\n- **Progressive disclosure**: Use `\\pause` or `\\onslide` for complex method slides.\n- **De-AI polish**: Remove watch words from all slide text and talk script.\n- **Do NOT hallucinate citations.** Reference only papers cited in the paper.\n- **Opening hook matters**: Never start with \"In this paper, we...\" — start with the problem or a provocative question.\n- **Font size minimums**: Title ≥28pt, body ≥20pt, footnotes ≥14pt.\n- **Feishu notifications are optional.** If `~\u002F.claude\u002Ffeishu.json` exists, send notifications. If absent, skip.\n\n## Parameter Pass-Through\n\n```\n\u002Fpaper-slides \"paper\u002F\" — talk_type: oral, venue: ICML, minutes: 20, aspect: 4:3, notes: false\n```\n\n| Parameter | Default | Description |\n|-----------|---------|-------------|\n| `venue` | NeurIPS | Conference for color scheme |\n| `talk_type` | spotlight | oral\u002Fspotlight\u002Fposter-talk\u002Finvited |\n| `minutes` | 15 | Talk duration |\n| `aspect` | 16:9 | Aspect ratio (16:9 \u002F 4:3) |\n| `notes` | true | Generate speaker notes |\n| `engine` | pdflatex | LaTeX engine |\n| `auto proceed` | false | Skip checkpoints |\n",{"name":5,"description":7,"allowed-tools":24,"argument-hint":25},"Bash(*), Read, Write, Edit, Grep, Glob, mcp__codex__codex, mcp__codex__codex-reply","[paper-directory-or-talk-length] [— style-ref: \u003Csource>]","main","18193d71ea3441d7c033dbf195b8a7c75ac96f75","packages\u002Fskill\u002Fskill-science\u002Fassets\u002Fsources\u002Faris\u002Fskills\u002Fpaper-slides",0,"https:\u002F\u002Fgithub.com\u002FCZX-LBJ\u002Fdeepseek-science","https:\u002F\u002Fgithub.com\u002FCZX-LBJ","一个没事喜欢随便写点代码的小菜鸡\r\n","2026-10-22 09:48:04",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add CZX-LBJ\u002Fdeepseek-science --skill paper-slides",{"tool":40,"command":41},"bunx","bunx skills add CZX-LBJ\u002Fdeepseek-science --skill paper-slides",{"tool":43,"command":44},"pnpm","pnpm dlx skills add CZX-LBJ\u002Fdeepseek-science --skill paper-slides","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FCZX-LBJ\u002Fdeepseek-science\u002Fblob\u002Fmain\u002Fpackages\u002Fskill\u002Fskill-science\u002Fassets\u002Fsources\u002Faris\u002Fskills\u002Fpaper-slides\u002FSKILL.md",true,[],[49,56,63,70,77,84],{"skillNo":50,"name":51,"nameZh":6,"description":52,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":53,"categories":54,"keywords":55,"publishedAt":19},"01M7FEEKSNMRMKW6JEK93PFDFY","bio-free-energy-calculations","Performs alchemical free-energy calculations including relative binding free energy (RBFE \u002F FEP+) and absolute binding free energy (ABFE) via OpenFE, FEP+, GROMACS, AMBER pmemd, and OpenMM with explicit lambda scheduling, soft-core potentials, MBAR\u002FBAR analysis, cycle-closure validation, and protocol-appropriate enhanced sampling. Compares ML alternatives (Boltz-2 affinity, DeepDock). Use when ranking analogs by binding affinity beyond docking accuracy, performing prospective lead optimization, or validating SAR predictions.","https:\u002F\u002Fgithub.com\u002FCZX-LBJ\u002Fdeepseek-science\u002Fblob\u002Fmain\u002Fpackages\u002Fskill\u002Fskill-science\u002Fassets\u002Fsources\u002Fbioskills\u002Fchemoinformatics\u002Ffree-energy-calculations\u002FSKILL.md",[],[],{"skillNo":57,"name":58,"nameZh":6,"description":59,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":19},"01M7FEE5CTE5QK55CEMXNNVWY9","bio-conformer-generation","Generates 3D conformer ensembles using RDKit ETKDGv3 with knowledge-enhanced distance geometry, MMFF94\u002FUFF force-field optimization, CREST + GFN2-xTB semi-empirical refinement, and macrocycle-aware torsion preferences. Provides explicit decision rules for single vs ensemble conformer use, RMSD pruning, energy windows, conformer count, and force-field choice. Use when preparing 3D ligands for docking, generating descriptor input for 3D QSAR, or sampling macrocycle\u002Fpeptide conformational ensembles.","https:\u002F\u002Fgithub.com\u002FCZX-LBJ\u002Fdeepseek-science\u002Fblob\u002Fmain\u002Fpackages\u002Fskill\u002Fskill-science\u002Fassets\u002Fsources\u002Fbioskills\u002Fchemoinformatics\u002Fconformer-generation\u002FSKILL.md",[],[],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":19},"01M7FEDNGFEMFWD1MSEDR5H7CY","bio-admet-prediction","Predicts ADMET properties using ADMETlab 3.0 (119 platform features, including 77 prediction models with modeled-endpoint uncertainty), ADMET-AI, DeepChem MolNet, and chemprop D-MPNN with explicit handling of OECD QSAR principles, applicability domain assessment, calibration, hERG\u002FCYP\u002FAMES endpoints, and PAINS \u002F Lipinski \u002F Ro5 \u002F Veber \u002F BBB druglikeness filters. Use when filtering compounds for drug-likeness, prioritizing leads by predicted safety, or building an in-house ADMET QSAR model.","https:\u002F\u002Fgithub.com\u002FCZX-LBJ\u002Fdeepseek-science\u002Fblob\u002Fmain\u002Fpackages\u002Fskill\u002Fskill-science\u002Fassets\u002Fsources\u002Fbioskills\u002Fchemoinformatics\u002Fadmet-prediction\u002FSKILL.md",[],[],{"skillNo":71,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M7FED69GFB54FRF7AV5066DN","writing-systems-papers","Paragraph-level structural blueprint for 10-12 page systems papers targeting OSDI, SOSP, ASPLOS, NSDI, and EuroSys. Provides page allocation, paragraph templates, and writing patterns. Use when user says \"写系统论文\", \"systems paper structure\", \"OSDI paper\", \"SOSP paper\", or wants fine-grained structural guidance for a systems conference submission.","https:\u002F\u002Fgithub.com\u002FCZX-LBJ\u002Fdeepseek-science\u002Fblob\u002Fmain\u002Fpackages\u002Fskill\u002Fskill-science\u002Fassets\u002Fsources\u002Faris\u002Fskills\u002Fwriting-systems-papers\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M7FECVVDWZ0X0VEDTCE4B4DZ","wiki-enrich","Fill in the per-paper TODO sections of research-wiki\u002Fpapers\u002F\u003Cslug>.md pages that literature-ingest skills leave as bare scaffolds. Use when user says 'enrich wiki', 'fill paper TODOs', 'wiki body 補完', '把 paper 摘要寫進 wiki', 'research-wiki 自動填', or after a batch ingest that left papers\u002F as TODO scaffolds.","https:\u002F\u002Fgithub.com\u002FCZX-LBJ\u002Fdeepseek-science\u002Fblob\u002Fmain\u002Fpackages\u002Fskill\u002Fskill-science\u002Fassets\u002Fsources\u002Faris\u002Fskills\u002Fwiki-enrich\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":19},"01M7FECD4BCJ2MTSBDDSE3ZAJ5","web-debug-search","Search GitHub, Stack Exchange, Chinese technical communities, official documentation, and general developer web sources for software errors, compatibility problems, API usage questions, and real-world workarounds. Use for debugging and discovery only; results are not paper-citation evidence.","https:\u002F\u002Fgithub.com\u002FCZX-LBJ\u002Fdeepseek-science\u002Fblob\u002Fmain\u002Fpackages\u002Fskill\u002Fskill-science\u002Fassets\u002Fsources\u002Faris\u002Fskills\u002Fweb-debug-search\u002FSKILL.md",[],[]]