[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-92493":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":14,"stars7d":14,"stars30d":15,"stars90d":14,"forks30d":14,"starsTrendScore":14,"compositeScore":16,"rankGlobal":9,"rankLanguage":9,"license":17,"archived":18,"fork":18,"defaultBranch":19,"hasWiki":20,"hasPages":18,"topics":21,"createdAt":9,"pushedAt":9,"updatedAt":22,"readmeContent":23,"aiSummary":24,"trendingCount":14,"starSnapshotCount":14,"syncStatus":12,"lastSyncTime":25,"discoverSource":26},92493,"ai-paper2slide-skill","Leo1998-Lu\u002Fai-paper2slide-skill","Leo1998-Lu","Conference-Grade Paper-to-Slide Generation for AI Research.",null,"Python",129,2,1,0,75,48.93,"MIT License",false,"main",true,[],"2026-07-22 04:02:06","\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Fbanner.svg\" width=\"760\" alt=\"AI Paper2Slide Skill\">\n\u003C\u002Fp>\n\n\u003Ch2 align=\"center\">\u003Cb>Conference-Grade Paper-to-Slide Generation for AI Research\u003C\u002Fb>\u003C\u002Fh2>\n\n\u003Cp align=\"center\">\n  \u003Ci>Turn AI papers into polished conference slides — with LaTeX-only visuals, default PPTX + English DOCX delivery, and clean AI conference style.\u003C\u002Fi>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"#quick-start\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FQuick%20Start-Get%20Started-brightgreen?style=for-the-badge&logo=github&logoColor=white\" alt=\"Quick Start\">\u003C\u002Fa>\n  \u003Ca href=\"docs\u002Fi18n\u002FREADME_ZH.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F中文-阅读%20CN-red?style=for-the-badge&logo=github&logoColor=white\" alt=\"中文\">\u003C\u002Fa>\n  \u003Ca href=\"LICENSE\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-blue?style=for-the-badge&logo=github&logoColor=white\" alt=\"License: MIT\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"#overview\">Overview\u003C\u002Fa> ·\n  \u003Ca href=\"#quick-start\">Quick Start\u003C\u002Fa> ·\n  \u003Ca href=\"#features\">Features\u003C\u002Fa> ·\n  \u003Ca href=\"#citation\">Citation\u003C\u002Fa> ·\n  \u003Ca href=\"#star-history\">Star History\u003C\u002Fa>\n\u003C\u002Fp>\n\n---\n\n## Overview\n\n`ai-paper2slide-skill` is an open-source ChatGPT Skill for converting AI and scientific papers into **conference-quality slide packages**.\n\nIt is designed for papers submitted to or presented at top-tier venues such as **NeurIPS, ICML, ICLR, CVPR, ACL, KDD, WWW, AAAI, SIGIR, EMNLP, ECCV, ICCV, MICCAI**, and related AI, machine learning, and computer vision conferences.\n\nUnlike generic document-to-slide tools, this Skill focuses on two strict production guarantees:\n\n1. **LaTeX-only slide images**  \n   Any image inserted into the slide deck must come from the user-provided LaTeX source package. The Skill does not use PDF screenshots, web images, generated images, previous-conversation images, or external substitutes.\n\n2. **Default two-file delivery**  \n   Every full paper-to-slide run returns at least two user-facing files by default: a `.pptx` slide deck and a `.docx` English per-slide speaker script.\n\n---\n\n## Quick Start\n\n> [!IMPORTANT]\n> Provide the original LaTeX source package first if you want source-grounded figures in the slides.\n>\n> ```text\n> Upload the LaTeX source package, and I will generate a conference-style PPTX deck plus an English DOCX speaker script.\n> ```\n\n### Recommended input\n\n1. **Original LaTeX source package**  \n   Supported formats: `.zip`, `.tar`, `.tar.gz`\n\n2. **Compiled paper PDF**  \n   Used for text cross-checking, section order, and final paper layout reference.\n\n### Typical output\n\n- `paper2slide_deck.pptx`\n- `paper2slide_speaker_script.docx`\n- `visual_source_manifest.json`\n- `paper2slide_quality_report.md`\n\n---\n\n## Features\n\n### LaTeX-only visual provenance\n\nThe Skill prioritizes the original LaTeX source package and scans it for:\n\n- `\\includegraphics` paths\n- figure and table environments\n- captions\n- labels\n- surrounding explanatory text\n- architecture and method diagrams\n- experimental result tables\n- ablation and qualitative visualizations\n\nOnly visual assets found inside the user-provided LaTeX package are eligible for slide images.\n\n### Default PPTX + English DOCX output\n\nFor every full paper-to-slide request, the Skill produces by default:\n\n| Output File | Required | Description |\n|---|---:|---|\n| `paper2slide_deck.pptx` | Yes | A 16:9 PowerPoint presentation for the paper |\n| `paper2slide_speaker_script.docx` | Yes | English per-slide speaking script, approximately 10 minutes total by default |\n| `visual_source_manifest.json` | Recommended | Source trace of figures and tables from the LaTeX package |\n| `paper2slide_quality_report.md` | Recommended | Checks for LaTeX-only image compliance, visual-source accuracy, slide readability, and script timing |\n\n### Conference-ready slide design\n\nThe Skill guides ChatGPT to create a clean, presentation-ready `.pptx` deck with:\n\n- 16:9 widescreen layout\n- claim-based slide titles\n- sparse and readable text\n- source-grounded architecture and result visuals\n- strong method storytelling\n- clear experiment and ablation slides\n- final takeaway and impact slide\n\n### Configurable presentation language\n\nThe default speaker script is **English**. Users may still request different language modes for slide text, notes, or additional scripts:\n\n| Mode | Description |\n|---|---|\n| `English` | Default mode for international AI conference talks |\n| `Chinese` | Suitable for Chinese group meetings, thesis defenses, and internal research reports |\n| `Bilingual` | English slide titles with Chinese explanations, or Chinese slides with English technical terms |\n| `Custom language` | Any user-specified language depending on the venue or audience |\n\n---\n\n## Example Requests\n\n### International conference talk\n\n```text\nConvert this LaTeX paper package into a 10-minute NeurIPS-style presentation.\n\nPlease create:\n1. A 16:9 PowerPoint slide deck.\n2. A Word document with the English per-slide speaking script.\n3. A visual source manifest for all architecture figures and experimental result tables.\n4. A quality report checking LaTeX-only image compliance, figure\u002Ftable placement, and script timing.\n\nOnly use images that are included in the provided LaTeX package.\n```\n\n### Chinese group meeting\n\n```text\n请将这篇论文转换为一个 10 分钟中文组会汇报。\n\n请生成：\n1. 中文 PPT。\n2. 默认英文逐页演讲稿 Word 文档。\n3. 可选中文逐页演讲稿 Word 文档。\n4. 图表来源 manifest 和质量报告。\n\nSlide 中的所有图片必须只来自我上传的 LaTeX 源文件包。\n```\n\n---\n\n## Included Helper Scripts\n\n### `scripts\u002Finspect_latex_assets.py`\n\nScans a LaTeX project or archive and creates a visual source manifest.\n\n```bash\npython scripts\u002Finspect_latex_assets.py paper_source.zip \\\n  --output visual_source_manifest.json\n```\n\n### `scripts\u002Fvalidate_visual_sources.py`\n\nChecks whether a slide visual map only references valid figure and table IDs from the manifest.\n\n```bash\npython scripts\u002Fvalidate_visual_sources.py \\\n  visual_source_manifest.json \\\n  slide_visual_map.json \\\n  --strict-latex-images \\\n  --output paper2slide_quality_report.md\n```\n\n---\n\n## Suggested 10-Minute Talk Structure\n\n| Section | Suggested Time | Purpose |\n|---|---:|---|\n| Title and motivation | 0.5-1 min | Introduce the problem and why it matters |\n| Key challenge | 1 min | Explain the technical gap |\n| Main idea | 1 min | Present the core insight |\n| Method overview | 2 min | Explain the model architecture and pipeline |\n| Main results | 2 min | Show quantitative improvements |\n| Ablation and analysis | 1-1.5 min | Validate design choices |\n| Qualitative examples | 1 min | Provide intuitive evidence, only if source assets exist |\n| Conclusion | 0.5 min | Summarize contributions and impact |\n\n---\n\n## Skill Structure\n\n```text\nai-paper2slide-skill\u002F\n├── README.md\n├── docs\u002F\n│   └── i18n\u002F\n│       └── README_ZH.md\n├── LICENSE\n├── assets\u002F\n│   └── README.md\n├── references\u002F\n│   ├── ai_conference_style_guide.md\n│   ├── latex_visual_localization.md\n│   ├── quality_checklist.md\n│   └── speaker_script_guide.md\n└── scripts\u002F\n    ├── inspect_latex_assets.py\n    └── validate_visual_sources.py\n```\n\n---\n\n## Design Principles\n\n### 1. Provenance before aesthetics\n\nThe Skill verifies where every slide image comes from before designing the slide around it.\n\n### 2. LaTeX package as the only image source\n\nIf an image is not in the user-provided LaTeX source package, it does not go into the deck.\n\n### 3. PPTX and English DOCX by default\n\nA complete run should always return the PowerPoint deck and the English per-slide speaker script.\n\n### 4. Claims before decoration\n\nEach slide should communicate one research claim, supported by one source-grounded visual or concise evidence.\n\n### 5. Conference realism\n\nThe deck should look like a polished AI conference presentation, not a generic corporate template.\n\n---\n\n## Limitations\n\n- The Skill works best when the LaTeX source package is complete and well-organized.\n- PDF-only workflows cannot provide paper images under the default strict image policy.\n- Complex TikZ figures, rasterized tables, or missing assets may require manual checking.\n- The Skill provides a workflow and helper scripts; final artifact generation depends on the ChatGPT environment and available document\u002Fslide tools.\n\n---\n\n## Roadmap\n\nPotential future improvements include:\n\n- stricter LaTeX asset provenance checking\n- better table-to-slide summarization\n- built-in PPTX template themes\n- venue-specific slide styles\n- automatic timing calibration from generated scripts\n- poster-to-slide conversion\n- multilingual supplemental speaker scripts\n- bilingual slide templates\n- integration with open-source presentation agents\n\n---\n\n## Contributing\n\nContributions are welcome.\n\nUseful contribution areas include:\n\n- better LaTeX parsing\n- improved slide quality checks\n- additional conference style guides\n- PPTX template assets that do not include external images\n- evaluation examples\n- multilingual documentation\n- support for more paper formats\n\nPlease open an issue or pull request with a clear description of the proposed improvement.\n\n---\n\n## Citation\n\nIf this Skill helps your research presentation workflow, please consider citing or linking to the repository.\n\n```bibtex\n@misc{ai-paper2slide-skill,\n  title        = {AI Paper2Slide Skill: Conference-Grade Paper-to-Slide Generation for AI Research},\n  author       = {Zhixiang Lu},\n  year         = {2026},\n  howpublished = {\\url{https:\u002F\u002Fgithub.com\u002FLeo1998-Lu\u002Fai-paper2slide-skill}},\n  note         = {Open-source ChatGPT Skill for LaTeX-only, conference-style paper-to-slide generation.}\n}\n```\n\n---\n\n## Star History\n\n![Star History](.\u002Fstar_history.png)\n","这是一个面向AI研究者的学术论文转会议幻灯片工具，专为NeurIPS、ICML、CVPR等顶级AI会议场景优化。核心功能是基于用户提供的原始LaTeX源码包，严格提取其中的图表、公式与结构化内容，生成符合学术规范的PPTX幻灯片和英文DOCX讲稿，所有图像均源自LaTeX源文件，杜绝PDF截图或AI生成图。适用于AI领域研究人员快速准备会议报告、学术汇报或论文答辩材料。","2026-07-09 02:30:07","CREATED_QUERY"]