Skip to content
Back to search

yixueAIganhuo-PPT

Use when creating a complete editable medical academic PowerPoint deck from papers, PDFs, Figures, screenshots, or support documents, or when converting an existing visual slide deck/image PDF into editable PPTX. Handles task routing, medical source understanding, Figure inventory, PPT planning, GPT Image 2 slide generation through editppt image, page-worker reconstruction, and final PPTX assembly.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/snowmanzhuang--yixueaiganhuo-ppt/badge.svg)](/listings/snowmanzhuang--yixueaiganhuo-ppt)

HTML

<a href="/listings/snowmanzhuang--yixueaiganhuo-ppt"><img src="/api/listings/snowmanzhuang--yixueaiganhuo-ppt/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

手动安装

把本仓库放到 Codex skills 目录:

mkdir -p ~/.codex/skills
git clone https://github.com/snowmanzhuang/yixueAIganhuo-PPT.git ~/.codex/skills/yixueAIganhuo-PPT

进入仓库后安装依赖:

cd ~/.codex/skills/yixueAIganhuo-PPT
python3 -m pip install -r requirements.txt

如果你的环境使用虚拟环境或 `uv`,可以按自己的 Python 环境管理方式安装依赖。

让 Codex 帮你安装

如果你不想自己敲命令,可以直接对 Codex 说:

帮我把 https://github.com/snowmanzhuang/yixueAIganhuo-PPT 安装或更新到 ~/.codex/skills/yixueAIganhuo-PPT,然后进入这个目录,按 requirements.txt 安装 Python 依赖。

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

SKILL.md · Manifest · 303 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. ---
  2. name: yixueAIganhuo-PPT
  3. description: Use when creating a complete editable medical academic PowerPoint deck from papers, PDFs, Figures, screenshots, or support documents, or when converting an existing visual slide deck/image PDF into editable PPTX. Handles task routing, medical source understanding, Figure inventory, PPT planning, GPT Image 2 slide generation through editppt image, page-worker reconstruction, and final PPTX assembly.
  4. ---
  5. # yixueAIganhuo-PPT
  6. `yixueAIganhuo-PPT` is now the 唯一完整 skill for the end-to-end workflow:
  7. ```text
  8. medical sources
  9. -> 医学上游 worker
  10. -> ppt_plan.json and slide prompts
  11. -> 页面生成 worker via editppt image
  12. -> page worker editable reconstruction
  13. -> editppt run finalize
  14. -> final editable PPTX
  15. ```
  16. The 统一父 agent in this skill is a scheduler. It creates the task directory, maintains global state, dispatches workers, controls concurrency, collects structured results, and runs deterministic `editppt` commands. It does not personally absorb full PDF text, every Figure detail, every prompt, OCR details, or page-level `manifest.json` decisions into its context.
  17. Parent context budget: keep the parent thread limited to routing decisions, worker dispatch, state paths, validation summaries, and final reporting. Do not perform medical PPT planning in the parent thread. The parent must dispatch the 医学上游 worker before reading full source text and must keep structured artifacts only in its own context.
  18. Do not ask the user to approve sub-agent spawning. Once the skill is chosen and prerequisites are satisfied, sub-agent dispatch is part of the workflow. Do not ask for a reply such as a permission phrase; proceed with worker dispatch and report progress.
  19. ## References
  20. - `references/single-skill-prd.md`: product and architecture baseline for the merged single-skill workflow.
  21. - `references/workflow-overview.md`: upstream medical planning DAG.
  22. - `references/prompt-key-selection.md`: style key selection and per-slide prompt construction.
  23. - `references/style_selector/`: local HTML style selector page and 001-019 template/example preview assets.
  24. - `references/provider-config.md`: unified `editppt image` backend contract.
  25. - `references/medical-integrated-workflow.md`: parent orchestration, worker boundaries, concurrency, incremental page intake, and failure handling.
  26. - `references/cli-helper.md`: `editppt` install check and command manual.
  27. - `references/page-decision-tree.md`: page-worker object-source decisions.
  28. - `references/manifest-schema.md`: `editppt` run/page JSON contracts.
  29. - `scripts/start-style-selector.py`: launches the local style selector page from the skill bundle.
  30. - `scripts/classify-task-route.py`: lightweight helper for initial task-route classification.
  31. - `scripts/validate-slide-prompt.py`: full-authoring rich GPT Image 2 prompt gate before slide-image worker dispatch.
  32. - `scripts/validate-slide-ready.py`: full-authoring preflight gate before downstream conversion commands.
  33. - `prompts/medical-upstream-worker.md`: prompt template for medical planning workers.
  34. - `prompts/medical-slide-image-worker.md`: prompt template for single-slide image generation workers.
  35. - `prompts/page-worker.md`: prompt template for downstream editable page workers.
  36. ## Retained Capabilities
  37. The skill keeps the original medical creation value:
  38. - source reading and concise medical summary
  39. - PDF/Figure inventory and evidence Figure selection
  40. - target slide structure and narrative planning
  41. - `ppt_plan.json`
  42. - Chinese speaker notes
  43. - page dependencies
  44. - per-slide GPT Image prompt writing
  45. The skill also embeds the downstream editable reconstruction value:
  46. - `editppt` CLI
  47. - page worker dispatch
  48. - `manifest.json`
  49. - OCR/text hints
  50. - foreground image asset reconstruction
  51. - PPT native element reconstruction
  52. - page validation
  53. - final deck assembly through `editppt run finalize`
  54. ## Deprecated Components
  55. The old yixue text-overlay editable path is deprecated. Do not use legacy OCR/image-repair PPTX scripts as the current execution path. The authoritative editable path is now the embedded `editppt` page-worker workflow.
  56. ## Task Routing
  57. Before creating downstream jobs, inspect the user request and input files, then write `task_route` into `task_state.json`.
  58. Recommended helper:
  59. ```bash
  60. python yixueAIganhuo-PPT/scripts/classify-task-route.py \
  61. --user-request "<original user request>" \
  62. --input-summary "<brief first-page/overview/OCR/metadata notes>" \
  63. <input paths...>
  64. ```
  65. Use the helper output as structured guidance, then apply the rules below. If a PDF preview clearly shows title/authors/abstract/body/references, choose `FULL_AUTHORING_FROM_SOURCE` even when the user says "转 PPTX".
  66. | `task_route` | Use When | Execution |
  67. |---|---|---|
  68. | `FULL_AUTHORING_FROM_SOURCE` | Inputs are source material for a new medical PPT: research PDF, medical paper, report, article, manuscript, guideline, support document, Figure folder, chart image, screenshot evidence, or mixed reference assets. A PDF with title/authors/abstract/body/references or journal-style columns is source material even if the user says "转 PPTX". | Default route. Run the 医学上游 worker first, create `ppt_plan.json`, generate new 16:9 slide images through `editppt image`, then send generated slide images to page workers. Do not run `editppt prepare` on a research/material PDF as the first action. |
  69. | `DIRECT_VISUAL_CONVERSION` | The user explicitly wants an existing visual slide deck, screenshot set, exported image-only PPT, or already-designed page PDF converted to editable PPTX while preserving the current pages as-is. Clear visual signals include 16:9 slide pages, existing slide titles/layouts, or page screenshots meant to be reconstructed rather than interpreted as evidence. | Skip medical upstream authoring. Run direct visual conversion with `editppt prepare` on the visual input, then page workers and `editppt run finalize`. |
  70. | `AMBIGUOUS_INPUT` | A PDF/image set could plausibly be either source material or an existing visual slide deck after quick inspection. | Ask one concise clarification. Suggested wording: "这份 PDF 是制作医学 PPT 的资料,还是已有幻灯片/截图需要转可编辑 PPTX?" If inspection clearly shows a medical paper/source document, do not ask; choose `FULL_AUTHORING_FROM_SOURCE`. |
  71. Do not treat a research/material PDF as an existing visual slide deck merely because it is a PDF.
  72. ## Full Authoring Hard Gate
  73. For `FULL_AUTHORING_FROM_SOURCE`, the final PPTX is invalid unless every slide passes the complete chain:
  74. 1. `prompts/slideNN.txt` follows the rich prompt schema from `references/prompt-key-selection.md` and passes:
  75. ```bash
  76. python yixueAIganhuo-PPT/scripts/validate-slide-prompt.py \
  77. --prompt <task>/prompts/slideNN.txt \
  78. --slide-number <N>
  79. ```
  80. 2. `slides/slideNN.png` is generated by `editppt image generate/edit/batch` through the 页面生成 worker.
  81. 3. `slide_results/slideNN.json` has `status: "IMAGE_READY"`, `passed: true`, and `image_backend.tool_call` showing `editppt image generate/edit/batch`.
  82. 4. `scripts/validate-slide-ready.py` passes for every generated slide before `editppt prepare` or `editppt run add-page`.
  83. 5. Every page is reconstructed by a page worker, recorded by `editppt run record`, and assembled only by `editppt run finalize`.
  84. Do not use @oai/artifact-tool, python-pptx, HTML/SVG screenshots, local PIL composition, direct native PPT generation, legacy yixue scripts, or any presentation/PPTX authoring shortcut as a substitute final path. If workers or image backend are unavailable, stop and report the blocker; do not downgrade to a simpler PPTX path.
  85. ## Preflight Questions Gate
  86. Before dispatching the 医学上游 worker for `FULL_AUTHORING_FROM_SOURCE`, ask a concise preflight question block unless all required values are already known.
  87. Before showing the block, start the bundled style selector server from this skill and use the printed `点击打开:...` URL as `{style_selector_url}`. Do not automatically open the browser; present the URL in the question block so the user can click it when needed:
  88. ```bash
  89. python3 yixueAIganhuo-PPT/scripts/start-style-selector.py --background
  90. ```
  91. If the local server cannot start, use the printed `本地文件兜底:...` file URL instead. The launcher resolves paths relative to its own script location, so it works when the whole skill folder is copied to another computer or installed under another username.
  92. The block must start with: "请先回答下面几个问题:"
  93. Use numbered emoji labels only:
  94. ```text
  95. 请先回答下面几个问题:
  96. 1️⃣ 页数:这次生成多少页 PPT?
  97. 2️⃣ 风格:请打开风格选择页,选 001-019 后回复序号。
  98. {style_selector_url}
  99. 不选则默认:001 通用提示词风格。
  100. ```
  101. If no PaddleOCR-VL token is detected, append the third question:
  102. ```text
  103. 3️⃣ OCR:未检测到 PaddleOCR-VL token。
  104. 需要更准的文字识别,请打开这里获取 token:
  105. https://aistudio.baidu.com/account/accessToken
  106. 然后回复 token;也可以回复“跳过”继续。
  107. ```
  108. If a PaddleOCR-VL token is already detected, do not show the OCR prompt; ask only questions 1️⃣ and 2️⃣. The OCR token question is optional and must not block generation if the user chooses to skip.
  109. The page count remains a hard gate. Ask the user for the target slide count before dispatching the 医学上游 worker unless the request already gives an explicit count. Do not default to any page count. Do not offer preset page-count choices. Record the confirmed number in `task_state.json` as `target_slide_count_confirmed` and pass it to the upstream worker as `TARGET_SLIDE_COUNT`.
  110. The style selector is optional. If the user gives a valid selector (`001`-`019`, a matching style name fragment, or a style JSON filename/path), record it in `task_state.json` as `style_selector` and pass it to the upstream worker as `STYLE_SELECTOR`. If the user does not choose a style, record and pass `STYLE_SELECTOR=001`.
  111. If the user gives a range, ask for one concrete number. If the user refuses to decide, stop and ask for a numeric page count; do not silently choose a count and do not present preset options.
  112. ## Parent Workflow
  113. 1. Create one task directory, for example `output/yixueAIganhuo-PPT/<run_id>/`.
  114. 2. Classify the task with the Task Routing rules and write `task_route` to `task_state.json`.
  115. 3. If `task_route` is `DIRECT_VISUAL_CONVERSION`, follow Direct Visual Conversion Workflow below and skip the full-authoring steps 4-15.
  116. 4. Write `task_state.json` with run phase, worker ids, per-slide status, paths, retries, concurrency slots, `task_route`, `style_selector`, and optional `style_selector_url`.
  117. 5. Dispatch the 医学上游 worker using `prompts/medical-upstream-worker.md`. The parent passes `TARGET_SLIDE_COUNT` only after `target_slide_count_confirmed=true`, and passes `STYLE_SELECTOR` after the user chooses a style or the parent defaults it to `001`; it does not create the PPT outline or slide prompts itself.
  118. 6. Receive only structured upstream artifacts:
  119. - `source_inventory.json`
  120. - `figure_inventory.json`
  121. - `plans/ppt_plan.json`
  122. - `prompts/slideNN.txt`
  123. 7. Validate that `ppt_plan.json` has one entry per target slide, continuous slide numbers, speaker notes, asset bindings, prompt files, and page dependencies.
  124. 8. Validate every slide prompt before dispatching 页面生成 workers:
  125. ```bash
  126. python yixueAIganhuo-PPT/scripts/validate-slide-prompt.py \
  127. --prompt <task>/prompts/slideNN.txt \
  128. --slide-number <N>
  129. ```
  130. Do not dispatch a 页面生成 worker with a summary-style prompt, a prompt that only says "medical academic style", or a prompt that lacks layout geometry, evidence/asset bindings, negative constraints, and continuity rules.
  131. 9. Configure or verify the unified image backend:
  132. ```bash
  133. editppt setup
  134. editppt doctor
  135. ```
  136. Use `editppt config` only when API fallback credentials are needed.
  137. 10. Dispatch 页面生成 worker jobs:
  138. - slide 1 first
  139. - slide 2 after slide 1 exists, with slide 1 as style reference
  140. - slide 3+ after slide 2 exists, with slide 1 and slide 2 as style references
  141. 11. Every 页面生成 worker must call `editppt image generate/edit/batch`. It writes `slides/slideNN.png` and `slide_results/slideNN.json`.
  142. 12. Before the first generated slide enters downstream conversion, run the real slide image preflight:
  143. ```bash
  144. python yixueAIganhuo-PPT/scripts/validate-slide-ready.py \
  145. --slide <task>/slides/slide01.png \
  146. --result <task>/slide_results/slide01.json
  147. ```
  148. Do not call `editppt prepare` until this preflight passes. A valid full-authoring slide requires `slides/slideNN.png`, `slide_results/slideNN.json`, `status: "IMAGE_READY"`, `passed: true`, provenance from `editppt image generate/edit/batch`, 16:9 dimensions, and nonblank image content. Forbidden inputs include blank or white placeholder PNGs, pure-color or low-variance placeholders, synthetic `slides/sources/slideNN.png` files used only to initialize `editppt_run`, and any hand-written page manifest built against placeholder `source.png`.
  149. 13. As soon as the first generated page image passes preflight, create the conversion run:
  150. ```bash
  151. editppt prepare <task>/slides/slide01.png --job-dir <task>/editppt_run
  152. ```
  153. 14. As later slide images finish, run the same preflight for each slide before appending it:
  154. ```bash
  155. python yixueAIganhuo-PPT/scripts/validate-slide-ready.py \
  156. --slide <task>/slides/slideNN.png \
  157. --result <task>/slide_results/slideNN.json
  158. ```
  159. 15. Append only preflight-passed slide images:
  160. ```bash
  161. editppt run add-page <task>/editppt_run <task>/slides/slideNN.png --source-page <N>
  162. ```
  163. 16. Use `editppt run next <task>/editppt_run --json` to drive page conversion. For each suggested page:
  164. ```bash
  165. python yixueAIganhuo-PPT/scripts/build-page-worker-prompt.py \
  166. <task>/editppt_run \
  167. --page page_001 \
  168. --out <task>/editppt_run/pages/page_001/worker-prompt.md
  169. ```
  170. Spawn a real page worker, then record the dispatch:
  171. ```bash
  172. editppt run dispatch <task>/editppt_run \
  173. --page page_001 \
  174. --agent-id <worker-id> \
  175. --prompt-file <task>/editppt_run/pages/page_001/worker-prompt.md
  176. ```
  177. After the worker returns:
  178. ```bash
  179. editppt run record <task>/editppt_run --page page_001 --agent-id <worker-id>
  180. ```
  181. 17. When all pages are recorded:
  182. ```bash
  183. editppt run finalize <task>/editppt_run
  184. ```
  185. 18. Report `editable_output`, `image_only_output`, validation result, page count, warnings, and retries.
  186. ## Direct Visual Conversion Workflow
  187. Use this branch only for `DIRECT_VISUAL_CONVERSION`, where the input is an existing visual slide deck or already-designed page images/PDF that should be preserved as-is.
  188. 1. Create the conversion run directly from the visual input:
  189. ```bash
  190. editppt prepare <visual-input> --job-dir <task>/editppt_run
  191. ```
  192. 2. Use the same page-worker loop: `editppt run next`, `scripts/build-page-worker-prompt.py`, worker dispatch, `editppt run record`.
  193. 3. When every page is recorded, run `editppt run finalize <task>/editppt_run`.
  194. 4. Report `editable_output`, `image_only_output`, validation result, page count, warnings, and retries.
  195. ## Concurrency
  196. - The agent pool starts with one 医学上游 worker for source understanding, PPT planning, speaker notes, and prompt writing.
  197. - After upstream artifacts are valid, dispatch multiple 页面生成 worker jobs according to slide dependencies.
  198. - As soon as any slide image passes `scripts/validate-slide-ready.py`, dispatch a page worker for that page immediately; do not wait for every slide image to finish.
  199. - Once any slide image exists, reduce slide-image generation concurrency to 2 or less.
  200. - The agent pool gradually reallocates worker slots to page workers as more validated slide images arrive.
  201. - Reserve remaining capacity for downstream page workers.
  202. - Keep page-worker concurrency bounded by `page_jobs.json.max_concurrent_pages`.
  203. - Never dispatch two workers for the same slide/page.
  204. - Never retry an unchanged failed page. Read the failure file, fix root cause, then `editppt run reset` and dispatch a fresh worker.
  205. ## Worker Boundaries
  206. ### 医学上游 worker
  207. Owns source understanding, Figure inventory, `ppt_plan.json`, speaker notes, page dependencies, and slide prompt files. It does not generate PPTX files or page-level `manifest.json`.
  208. ### 页面生成 worker
  209. Owns exactly one generated slide image. It uses only `editppt image generate/edit/batch`, writes `slides/slideNN.png`, and returns `slide_results/slideNN.json`.
  210. ### page worker
  211. Owns exactly one `editppt_run/pages/page_NNN/` directory. It follows `prompts/page-worker.md`, `references/page-decision-tree.md`, and `references/manifest-schema.md`, then returns page conversion result paths.
  212. ## Required Final Output
  213. The final deliverables are both required:
  214. - final editable PPTX: object-level editable `.pptx` created by `editppt run finalize`
  215. - final image-only PPTX: a full-slide raster companion deck created from the generated slide images
  216. Both outputs must carry speaker notes from `notes_manifest.json` when notes are available. The final report must include `editable_output` and `image_only_output`.
README.md · README · 188 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # yixueAIganhuo-PPT
  2. [中文](README.md) | [English](README.en.md)
  3. 把医学论文、PDF、Figure、截图和参考资料自动整理成医学学术 PPT。
  4. 默认同时交付 **图片型 PPTX** 和 **对象级可编辑 PPTX**,并写入每页中文 speaker notes。
  5. ![yixueAIganhuo-PPT workflow](docs/workflow-overview.webp)
  6. ## 这个项目能做什么
  7. `yixueAIganhuo-PPT` 是一个面向 Codex 的医学 PPT 自动化 skill。它把医学材料理解、幻灯片视觉生成和可编辑 PPT 重建拆成三段:
  8. 1. **父 agent 编排上游**:读取论文、PDF、Figure、截图,完成医学理解、Figure 盘点、PPT 结构规划和每页讲稿备注。
  9. 2. **中游 GPT Image 2 生成图片型 PPT**:按页生成统一风格的医学学术 slide image,并组装图片型 PPTX。
  10. 3. **下游 page worker 重建可编辑 PPT**:把每页图像拆解成文字、图片资产、线条、卡片、表格、坐标轴和 PPT 原生 shape,最终生成可编辑 PPTX。
  11. 最终默认交付:
  12. - `editable PPTX`:对象级可编辑,文字可改,图片、卡片、线条和图表元素可调整。
  13. - `image-only PPTX`:整页图片版本,适合分享、归档和兜底展示。
  14. - `speaker notes`:每页默认生成中文讲稿备注,不需要额外选择。
  15. ## 效果展示:图片型 PPTX vs 可编辑 PPTX
  16. 左侧是图片型 PPTX,右侧是可编辑 PPTX。右侧截图中可以看到 PowerPoint 全选后的文本框、图片框和 shape 边界。
  17. ![slide01 editable comparison](docs/editable-comparison-slide01.webp)
  18. ![slide05 editable comparison](docs/editable-comparison-slide05.webp)
  19. ![slide18 editable comparison](docs/editable-comparison-slide18.webp)
  20. ![slide22 editable comparison](docs/editable-comparison-slide22.webp)
  21. ## 19 种风格可选
  22. 生成前可以选择 001-019 号风格。不选择时默认使用 `001 通用医学汇报PPT风格`。
  23. ![19 styles template overview](docs/style-selector-19-template-overview.webp)
  24. ## 适合哪些场景
  25. - 📖 医学文献精读 PPT
  26. - 🧪 SCI 论文组会汇报
  27. - 🏥 科室教学课件
  28. - 🩺 病例讨论和专题学习
  29. - 📚 指南、综述、共识类内容整理
  30. - 🖼️ Figure-heavy 的医学研究汇报
  31. ## 环境要求
  32. ### 1. Codex
  33. 本项目作为 Codex skill 使用,需要在 Codex 环境中运行。推荐使用 **GPT Pro** 会员完成完整流程。
  34. Plus 会员也可以使用,但 token 额度可能更适合**只生成图片型 PPTX**;如果要完整跑完医学上游规划、GPT Image 2 按页生成、下游 page worker 可编辑重建,可能会遇到 token 或任务长度限制。
  35. ### 2. PaddleOCR-VL API token
  36. 下游可编辑重建需要更准确地识别页面文字,建议提前准备 PaddleOCR-VL token。
  37. 通过 https://aistudio.baidu.com/account/accessToken 获取。PaddleOCR-VL 目前提供每日 2 万张免费处理额度。
  38. ## 安装
  39. ### 手动安装
  40. 把本仓库放到 Codex skills 目录:
  41. ```bash
  42. mkdir -p ~/.codex/skills
  43. git clone https://github.com/snowmanzhuang/yixueAIganhuo-PPT.git ~/.codex/skills/yixueAIganhuo-PPT
  44. ```
  45. 进入仓库后安装依赖:
  46. ```bash
  47. cd ~/.codex/skills/yixueAIganhuo-PPT
  48. python3 -m pip install -r requirements.txt
  49. ```
  50. 如果你的环境使用虚拟环境或 `uv`,可以按自己的 Python 环境管理方式安装依赖。
  51. ### 让 Codex 帮你安装
  52. 如果你不想自己敲命令,可以直接对 Codex 说:
  53. ```text
  54. 帮我把 https://github.com/snowmanzhuang/yixueAIganhuo-PPT 安装或更新到 ~/.codex/skills/yixueAIganhuo-PPT,然后进入这个目录,按 requirements.txt 安装 Python 依赖。
  55. ```
  56. ## 快速开始
  57. 在 Codex 中直接提出任务,例如:
  58. ```text
  59. 请使用 yixueAIganhuo-PPT skill,把 ./paper.pdf 做成 12 页中文医学汇报 PPT。
  60. ```
  61. 也可以提供 Figure 文件夹、截图、已有参考资料:
  62. ```text
  63. 请使用 yixueAIganhuo-PPT skill,基于 ./paper.pdf 和 ./figures/ 生成 15 页中文课题组汇报 PPT,风格 009。
  64. ```
  65. 如果没有提前说明页数和风格,skill 会先询问:
  66. ```text
  67. 请先回答下面几个问题:
  68. 1️⃣ 页数:这次生成多少页 PPT?
  69. 2️⃣ 风格:请打开风格选择页,选 001-019 后回复序号。
  70. 不选则默认:001 通用提示词风格。
  71. 3️⃣ OCR:未检测到 PaddleOCR-VL token。
  72. 需要更准的文字识别,请打开这里获取 token:
  73. https://aistudio.baidu.com/account/accessToken
  74. 然后回复 token;也可以回复“跳过”继续。
  75. ```
  76. 页数是必须确认的;风格不选则默认 `001`。speaker notes 默认生成,不需要单独选择。
  77. ## 输入材料
  78. 支持输入:
  79. - 📄 医学论文 PDF
  80. - 📚 指南、综述、共识、报告类 PDF
  81. - 🖼️ Figure 图片文件夹
  82. - 🔬 实验结果截图、表格截图、影像截图
  83. - 📊 已有 PPT、图片型 PPT 或导出的 slide images
  84. - 🎯 用户补充的目标受众、汇报场景、页数和风格编号
  85. 🧠 如果输入是论文或资料,skill 会走完整创作流程:医学理解、Figure 盘点、PPT 规划、按页生成、可编辑重建。
  86. 🔁 如果输入已经是现成的 slide 截图或图片型 PPT,skill 会走直接视觉转换流程,尽量保留原页面并转成可编辑 PPTX。
  87. ## 输出结构
  88. 一次完整运行通常会生成类似结构:
  89. ```text
  90. output/yixueAIganhuo-PPT/<run_id>/
  91. ├── plans/
  92. │ └── ppt_plan.json
  93. ├── prompts/
  94. │ └── slideNN.txt
  95. ├── slides/
  96. │ └── slideNN.png
  97. ├── slide_results/
  98. │ └── slideNN.json
  99. ├── editppt_run/
  100. │ ├── pages/
  101. │ ├── manifest.json
  102. │ └── final outputs
  103. ├── editable.pptx
  104. ├── image-only.pptx
  105. └── notes_manifest.json
  106. ```
  107. 实际文件名会根据运行目录和 `editppt` 输出略有差异,但最终报告会明确列出:
  108. - `editable_output`
  109. - `image_only_output`
  110. - 页数
  111. - speaker notes 状态
  112. - 验证结果
  113. - warning / retry 信息
  114. ## 注意事项
  115. - 医学内容应由使用者复核,尤其是诊断、治疗、指南推荐和统计结论。
  116. - 原始证据图应来自用户提供的论文、Figure、截图或可信资料。
  117. - 可编辑 PPTX 是结构化重建结果,不保证 100% 等同人工精修排版。
  118. - 极复杂表格、密集多图页和特殊图表可能需要人工局部微调。
  119. - Plus 会员可能更适合先生成图片型 PPTX;完整可编辑重建推荐 GPT Pro。
  120. ## 开源协议
  121. 详见 [LICENSE](LICENSE)。
  122. ## 联系
  123. <img src="docs/yixue-ai-ganhuo-wechat-qrcode.jpg" alt="yixue AI ganhuo QR code" width="50%">
  124. ## 致谢
  125. 本项目的 page worker 大部分参考了 [image-to-editable-ppt-skill](https://github.com/ningzimu/image-to-editable-ppt-skill),感谢原项目作者的开源工作。

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.