Skip to content
Back to search

PPTAgent

An Agentic Framework for Reflective PowerPoint Generation

Safety scan

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

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/icip-cas/PPTAgent with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/icip-cas/PPTAgent'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/icip-cas--pptagent/badge.svg)](/listings/icip-cas--pptagent)

HTML

<a href="/listings/icip-cas--pptagent"><img src="/api/listings/icip-cas--pptagent/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.

**Requirements:** Claude Code, Codex CLI, or OpenCode; Linux (including WSL) or macOS; [uv](https://docs.astral.sh/uv/getting-started/installation/); npm; and LibreOffice available as `libreoffice` on PATH. On macOS, also install Google Chrome for the converter.

1. Install the runtime

On Debian/Ubuntu, install the system dependencies first:

sudo apt-get install npm libreoffice-impress

On macOS, install the system dependencies with Homebrew:

brew install node
brew install --cask libreoffice google-chrome
ln -sf "$(command -v soffice)" "$(brew --prefix)/bin/libreoffice"

Clone the repository and install the skill's dependencies. If you already have a checkout, start from its `skills/pptagent/` directory:

git clone https://github.com/icip-cas/PPTAgent.git
cd PPTAgent/skills/pptagent

uv venv --python 3.12 .venv
uv pip install --python .venv/bin/python -r requirements.txt
.venv/bin/python -m playwright install --with-deps chromium

2. Register with your coding agent

Run the command for your client from `skills/pptagent/`.

**Claude Code**

.venv/bin/python scripts/install.py --client claude

**Codex CLI**

.venv/bin/python scripts/install.py --client codex

**OpenCode**

.venv/bin/python scripts/install.py --client opencode

The installer prepares Node dependencies and registers the skill for the selected client. Keep the repository in place, then check the installation:

.venv/bin/python scripts/pptagent.py doctor

For OpenCode visual MCP configuration and the complete workflow, continue with the [OpenCode setup guide](skills/pptagent/README.md#opencode-setup).

<a id="quick-start"></a>

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.

README.md · README · 406 lines

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

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. <h1 align="center">
  2. <img src="resource/pptagent.png" width="300" alt="PPTAgent">
  3. </h1>
  4. <p align="center">
  5. <a href="https://atomgit.com/icip-cas/PPTAgent">
  6. <img src="https://atomgit.com/icip-cas/PPTAgent/star/new_badge.svg" alt="AtomGit Stars">
  7. </a>
  8. </p>
  9. > [!TIP]
  10. > **PPTAgent Skill for Claude Code, Codex & OpenCode Now Available!**
  11. >
  12. > Create, revise, and visually review editable PowerPoint decks with your coding agent.
  13. >
  14. > **[Install the Skill →](#install-skill)** · **[Use Atria Dawn Preview →](#quick-start)**
  15. >
  16. > **Free Token Plan:** [Discovery](https://discovery-home.intern-ai.org.cn/) · [Atria API](https://api.atria-asi.ai/)
  17. >
  18. > You can use a multimodal vision model as a multimodal reviewer and apply for an API token through the [Duanyan (端砚) Token Plan](https://discovery.intern-ai.org.cn/token-plan/home?tabIndex=1). We recommend `deepseek-v4-flash-vision` as the multimodal model. [Setup guide →](skills/pptagent/README.md#visual-review)
  19. > [!IMPORTANT]
  20. > **Looking for the previous runtime or reproducing the research papers?** Use
  21. > these pinned versions instead of the current development branch:
  22. >
  23. > - **PPTAgent (EMNLP 2025):** [paper](https://arxiv.org/abs/2501.03936) · [code at v0.2.0](https://github.com/icip-cas/PPTAgent/tree/v0.2.0)
  24. > - **DeepPresenter (ACL 2026):** [paper](https://arxiv.org/abs/2602.22839) · [code at v1.1.38](https://github.com/icip-cas/PPTAgent/tree/v1.1.38)
  25. <p align="center">
  26. <img src="resource/pptagent-demo.gif" width="50%" alt="PPTAgent demo">
  27. </p>
  28. ## 📅 News
  29. - **[2026/09]** 🚀 Introducing **Atria Dawn Preview**, a new agentic model jointly released by Shanghai AI Laboratory, Fudan University, the Institute of Software (Chinese Academy of Sciences), Renmin University of China, the Institute of Automation (Chinese Academy of Sciences), and East China Normal University. **Claim a generous free Token Plan:** [Discovery](https://discovery-home.intern-ai.org.cn/) · [Atria](https://api.atria-asi.ai/). [Use Atria with PPTAgent Skill →](skills/pptagent/README.md#try-atria)
  30. - **[2026/09]** 🧩 Released **[PPTAgent Skill](skills/pptagent/README.md)** for **Claude Code, Codex & OpenCode** — create, visually review, and export editable PowerPoint decks with your coding agent. [Get started →](skills/pptagent/README.md#quick-start)
  31. - **[2026/04]** 🎉 [DeepPresenter](https://arxiv.org/abs/2602.22839) accepted to **ACL 2026**!
  32. - **[2026/03]** 🤗 We released fine-tuned models and taskset on [Hugging Face](https://huggingface.co/collections/ICIP/deeppresenter).
  33. - **[2026/01]** 🆕 Freeform & template generation now support PPTX export and offline mode. Context management added to prevent context overflow.
  34. - **[2025/12]** 🔥 Released **DeepPresenter** codebase with major upgrades — Deep Research Integration, Free-Form Visual Design, Autonomous Asset Creation, Text-to-Image Generation, and an Agent Environment with sandbox & 20+ tools.
  35. - **[2025/09]** 🛠️ MCP server support added — see the [v1.1.37 documentation](https://github.com/icip-cas/PPTAgent/blob/v1.1.37/pptagent/DOC.md#mcp-server-) for details.
  36. - **[2025/08]** 🎉 [PPTAgent](https://arxiv.org/abs/2501.03936) accepted to **EMNLP 2025**!
  37. - **[2025/05]** ⭐ Reached **1,000 stars** on GitHub!
  38. - **[2025/01]** 🔓 Open-sourced the PPTAgent codebase.
  39. <a id="install-skill"></a>
  40. ## Install PPTAgent Skill 🧩
  41. **Requirements:** Claude Code, Codex CLI, or OpenCode; Linux (including WSL) or macOS; [uv](https://docs.astral.sh/uv/getting-started/installation/); npm; and LibreOffice available as `libreoffice` on PATH. On macOS, also install Google Chrome for the converter.
  42. ### 1. Install the runtime
  43. On Debian/Ubuntu, install the system dependencies first:
  44. ```bash
  45. sudo apt-get install npm libreoffice-impress
  46. ```
  47. On macOS, install the system dependencies with Homebrew:
  48. ```bash
  49. brew install node
  50. brew install --cask libreoffice google-chrome
  51. ln -sf "$(command -v soffice)" "$(brew --prefix)/bin/libreoffice"
  52. ```
  53. Clone the repository and install the skill's dependencies. If you already have a checkout, start from its `skills/pptagent/` directory:
  54. ```bash
  55. git clone https://github.com/icip-cas/PPTAgent.git
  56. cd PPTAgent/skills/pptagent
  57. uv venv --python 3.12 .venv
  58. uv pip install --python .venv/bin/python -r requirements.txt
  59. .venv/bin/python -m playwright install --with-deps chromium
  60. ```
  61. ### 2. Register with your coding agent
  62. Run the command for your client from `skills/pptagent/`.
  63. **Claude Code**
  64. ```bash
  65. .venv/bin/python scripts/install.py --client claude
  66. ```
  67. **Codex CLI**
  68. ```bash
  69. .venv/bin/python scripts/install.py --client codex
  70. ```
  71. **OpenCode**
  72. ```bash
  73. .venv/bin/python scripts/install.py --client opencode
  74. ```
  75. The installer prepares Node dependencies and registers the skill for the selected client. Keep the repository in place, then check the installation:
  76. ```bash
  77. .venv/bin/python scripts/pptagent.py doctor
  78. ```
  79. For OpenCode visual MCP configuration and the complete workflow, continue with
  80. the [OpenCode setup guide](skills/pptagent/README.md#opencode-setup).
  81. <a id="quick-start"></a>
  82. ## Quick Start with Atria Dawn Preview 🚀
  83. This example uses **Atria Dawn Preview** to write and revise the slides, with an external visual model to review them. Complete the installation above, then create an API key in the [Atria console](https://api.atria-asi.ai/console/keys).
  84. ### 1. Set up text-mode visual review
  85. You can use a multimodal vision model as a multimodal reviewer and apply for an API token through the [Duanyan (端砚) Token Plan](https://discovery.intern-ai.org.cn/token-plan/home?tabIndex=1). We recommend `deepseek-v4-flash-vision` as the multimodal model.
  86. In `skills/pptagent/config.yaml`, use the example below and replace `base_url` with the OpenAI-compatible API base URL shown in your Duanyan console:
  87. ```yaml
  88. mode: text
  89. visual:
  90. base_url: "<OpenAI-compatible API base URL from the Duanyan console>"
  91. model: "deepseek-v4-flash-vision"
  92. api_key_env: VISUAL_API_KEY
  93. timeout_seconds: 300
  94. delivery:
  95. mode: strict
  96. ```
  97. Save your Duanyan API key in `skills/pptagent/.env`:
  98. ```dotenv
  99. VISUAL_API_KEY=<your-duanyan-api-key>
  100. ```
  101. Atria writes the slides through the text workflow; `deepseek-v4-flash-vision` reviews the rendered images. The skill appends `/chat/completions` to `base_url`; use the API prefix from the console, not the Token Plan webpage URL. With an image-capable host model, you can use `mode: multimodal` instead. See [visual review configuration](skills/pptagent/README.md#visual-review) for details.
  102. ### 2. Launch your coding agent with Atria
  103. Set your Atria key and open a separate task folder:
  104. ```bash
  105. export ATRIA_API_KEY="<your-atria-api-key>"
  106. mkdir -p ~/pptagent-demo
  107. cd ~/pptagent-demo
  108. ```
  109. Choose the client you registered during installation.
  110. **Claude Code** — launch through Atria's Messages API:
  111. ```bash
  112. ANTHROPIC_BASE_URL=https://api.atria-asi.ai \
  113. ANTHROPIC_AUTH_TOKEN="$ATRIA_API_KEY" \
  114. claude --model Atria-Dawn-Preview
  115. ```
  116. **Codex CLI** — merge the following into `~/.codex/config.toml`. Keep `model` and `model_provider` at the top level, before any section headers, and update existing entries rather than duplicating them:
  117. ```toml
  118. model = "Atria-Dawn-Preview"
  119. model_provider = "atria"
  120. [model_providers.atria]
  121. name = "Atria"
  122. base_url = "https://api.atria-asi.ai/v1"
  123. env_key = "ATRIA_API_KEY"
  124. wire_api = "responses"
  125. ```
  126. Then run `codex` from the same terminal. It uses Atria's Responses API and reads `ATRIA_API_KEY` from your environment.
  127. **OpenCode** — configure Atria through your normal OpenCode provider settings,
  128. then follow the [OpenCode setup guide](skills/pptagent/README.md#opencode-setup)
  129. to connect the visual review tools.
  130. ### 3. Create your first presentation
  131. Send this request in your Atria-powered session:
  132. ```text
  133. Use the pptagent skill to create a 6-slide presentation about how our
  134. engineering team can adopt AI coding assistants. Use a clean 16:9 layout.
  135. Render and visually review the slides and the exported deck, then deliver
  136. an editable answer.pptx.
  137. ```
  138. You can also invoke the skill explicitly with `/pptagent` in Claude Code or `$pptagent` in Codex. In OpenCode, ask it to use the `pptagent` skill. Continue in the same session to revise the deck:
  139. ```text
  140. Turn slide 3 into a workflow diagram and shorten the recommendations.
  141. Keep the deck at 6 slides, rebuild it, and review the updated PPTX.
  142. ```
  143. Your task folder keeps **`answer.pptx`**, editable HTML sources, previews, and the review report. See the [Skill guide](skills/pptagent/README.md) for details and [optional MinerU and search tools](skills/pptagent/README.md#configuration) for richer source material.
  144. ## Contributors 🌟
  145. <table>
  146. <tr>
  147. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  148. <a href=https://github.com/Force1ess>
  149. <img src=https://avatars.githubusercontent.com/u/72636351?v=4 width="80;" alt=Force1ess/>
  150. <br />
  151. <sub style="font-size:14px"><b>Force1ess</b></sub>
  152. </a>
  153. </td>
  154. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  155. <a href=https://github.com/Puellaquae>
  156. <img src=https://avatars.githubusercontent.com/u/22560343?v=4 width="80;" alt=Puelloc/>
  157. <br />
  158. <sub style="font-size:14px"><b>Puelloc</b></sub>
  159. </a>
  160. </td>
  161. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  162. <a href=https://github.com/imHuZijian>
  163. <img src=https://avatars.githubusercontent.com/u/97173940?v=4 width="80;" alt=BrandonHu/>
  164. <br />
  165. <sub style="font-size:14px"><b>BrandonHu</b></sub>
  166. </a>
  167. </td>
  168. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  169. <a href=https://github.com/hysyyds>
  170. <img src=https://avatars.githubusercontent.com/u/80150669?v=4 width="80;" alt=hongyan/>
  171. <br />
  172. <sub style="font-size:14px"><b>hongyan</b></sub>
  173. </a>
  174. </td>
  175. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  176. <a href=https://github.com/Dnoob>
  177. <img src=https://avatars.githubusercontent.com/u/92987618?v=4 width="80;" alt=Dnoob/>
  178. <br />
  179. <sub style="font-size:14px"><b>Dnoob</b></sub>
  180. </a>
  181. </td>
  182. </tr>
  183. <tr>
  184. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  185. <a href=https://github.com/Sadahlu>
  186. <img src=https://avatars.githubusercontent.com/u/126563707?v=4 width="80;" alt=Sadahlu/>
  187. <br />
  188. <sub style="font-size:14px"><b>Sadahlu</b></sub>
  189. </a>
  190. </td>
  191. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  192. <a href=https://github.com/lnennnn>
  193. <img src=https://avatars.githubusercontent.com/u/124434018?v=4 width="80;" alt=lnennnn/>
  194. <br />
  195. <sub style="font-size:14px"><b>lnennnn</b></sub>
  196. </a>
  197. </td>
  198. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  199. <a href=https://github.com/KurisuMakiseSame>
  200. <img src=https://avatars.githubusercontent.com/u/168447425?v=4 width="80;" alt=KurisuMakiseSame/>
  201. <br />
  202. <sub style="font-size:14px"><b>KurisuMakiseSame</b></sub>
  203. </a>
  204. </td>
  205. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  206. <a href=https://github.com/RheagalFire>
  207. <img src=https://avatars.githubusercontent.com/u/60213893?v=4 width="80;" alt=Aarish Alam/>
  208. <br />
  209. <sub style="font-size:14px"><b>Aarish Alam</b></sub>
  210. </a>
  211. </td>
  212. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  213. <a href=https://github.com/Angelenx>
  214. <img src=https://avatars.githubusercontent.com/u/39873863?v=4 width="80;" alt=Angelen/>
  215. <br />
  216. <sub style="font-size:14px"><b>Angelen</b></sub>
  217. </a>
  218. </td>
  219. </tr>
  220. <tr>
  221. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  222. <a href=https://github.com/kylooh>
  223. <img src=https://avatars.githubusercontent.com/u/26456650?v=4 width="80;" alt=Eliot White/>
  224. <br />
  225. <sub style="font-size:14px"><b>Eliot White</b></sub>
  226. </a>
  227. </td>
  228. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  229. <a href=https://github.com/EvolvedGhost>
  230. <img src=https://avatars.githubusercontent.com/u/92856393?v=4 width="80;" alt=EvolvedGhost/>
  231. <br />
  232. <sub style="font-size:14px"><b>EvolvedGhost</b></sub>
  233. </a>
  234. </td>
  235. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  236. <a href=https://github.com/ISCAS-zwl>
  237. <img src=https://avatars.githubusercontent.com/u/179820048?v=4 width="80;" alt=ISCAS-zwl/>
  238. <br />
  239. <sub style="font-size:14px"><b>ISCAS-zwl</b></sub>
  240. </a>
  241. </td>
  242. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  243. <a href=https://github.com/James4Ever0>
  244. <img src=https://avatars.githubusercontent.com/u/103997068?v=4 width="80;" alt=白雨 | James Brown/>
  245. <br />
  246. <sub style="font-size:14px"><b>白雨 | James Brown</b></sub>
  247. </a>
  248. </td>
  249. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  250. <a href=https://github.com/LasRuinasCirculares>
  251. <img src=https://avatars.githubusercontent.com/u/119716645?v=4 width="80;" alt=JunZhang/>
  252. <br />
  253. <sub style="font-size:14px"><b>JunZhang</b></sub>
  254. </a>
  255. </td>
  256. </tr>
  257. <tr>
  258. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  259. <a href=https://github.com/openaitx-system>
  260. <img src=https://avatars.githubusercontent.com/u/215529505?v=4 width="80;" alt=Open AI Tx/>
  261. <br />
  262. <sub style="font-size:14px"><b>Open AI Tx</b></sub>
  263. </a>
  264. </td>
  265. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  266. <a href=https://github.com/haosenwang1018>
  267. <img src=https://avatars.githubusercontent.com/u/167664334?v=4 width="80;" alt=Sense_wang/>
  268. <br />
  269. <sub style="font-size:14px"><b>Sense_wang</b></sub>
  270. </a>
  271. </td>
  272. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  273. <a href=https://github.com/DeJeune>
  274. <img src=https://avatars.githubusercontent.com/u/67425183?v=4 width="80;" alt=SuYao/>
  275. <br />
  276. <sub style="font-size:14px"><b>SuYao</b></sub>
  277. </a>
  278. </td>
  279. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  280. <a href=https://github.com/JiwaniZakir>
  281. <img src=https://avatars.githubusercontent.com/u/108548454?v=4 width="80;" alt=Zakir Jiwani/>
  282. <br />
  283. <sub style="font-size:14px"><b>Zakir Jiwani</b></sub>
  284. </a>
  285. </td>
  286. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  287. <a href=https://github.com/Dormiveglia-elf>
  288. <img src=https://avatars.githubusercontent.com/u/81767213?v=4 width="80;" alt=Zhenyu/>
  289. <br />
  290. <sub style="font-size:14px"><b>Zhenyu</b></sub>
  291. </a>
  292. </td>
  293. </tr>
  294. <tr>
  295. <td align="center" style="word-wrap: break-word; width: 120.0; height: 120.0">
  296. <a href=https://github.com/wangzh12023>
  297. <img src=https://avatars.githubusercontent.com/u/146173014?v=4 width="80;" alt=Zihan Wang/>
  298. <br />
  299. <sub style="font-size:14px"><b>Zihan Wang</b></sub>
  300. </a>
  301. </td>
  302. </tr>
  303. </table>
  304. [![Star History Chart](https://api.star-history.com/svg?repos=icip-cas/PPTAgent&type=Date)](https://star-history.com/#icip-cas/PPTAgent&Date)
  305. ## Citation 🙏
  306. If you find this project helpful, please use the following to cite it:
  307. ```bibtex
  308. @inproceedings{zheng-etal-2025-pptagent,
  309. title = "{PPTA}gent: Generating and Evaluating Presentations Beyond Text-to-Slides",
  310. author = "Zheng, Hao and
  311. Guan, Xinyan and
  312. Kong, Hao and
  313. Zhang, Wenkai and
  314. Zheng, Jia and
  315. Zhou, Weixiang and
  316. Lin, Hongyu and
  317. Lu, Yaojie and
  318. Han, Xianpei and
  319. Sun, Le",
  320. editor = "Christodoulopoulos, Christos and
  321. Chakraborty, Tanmoy and
  322. Rose, Carolyn and
  323. Peng, Violet",
  324. booktitle = "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing",
  325. month = nov,
  326. year = "2025",
  327. address = "Suzhou, China",
  328. publisher = "Association for Computational Linguistics",
  329. url = "https://aclanthology.org/2025.emnlp-main.728/",
  330. doi = "10.18653/v1/2025.emnlp-main.728",
  331. pages = "14413--14429",
  332. ISBN = "979-8-89176-332-6",
  333. abstract = "Automatically generating presentations from documents is a challenging task that requires accommodating content quality, visual appeal, and structural coherence. Existing methods primarily focus on improving and evaluating the content quality in isolation, overlooking visual appeal and structural coherence, which limits their practical applicability. To address these limitations, we propose PPTAgent, which comprehensively improves presentation generation through a two-stage, edit-based approach inspired by human workflows. PPTAgent first analyzes reference presentations to extract slide-level functional types and content schemas, then drafts an outline and iteratively generates editing actions based on selected reference slides to create new slides. To comprehensively evaluate the quality of generated presentations, we further introduce PPTEval, an evaluation framework that assesses presentations across three dimensions: Content, Design, and Coherence. Results demonstrate that PPTAgent significantly outperforms existing automatic presentation generation methods across all three dimensions."
  334. }
  335. @misc{zheng2026deeppresenterenvironmentgroundedreflectionagentic,
  336. title={DeepPresenter: Environment-Grounded Reflection for Agentic Presentation Generation},
  337. author={Hao Zheng and Guozhao Mo and Xinru Yan and Qianhao Yuan and Wenkai Zhang and Xuanang Chen and Yaojie Lu and Hongyu Lin and Xianpei Han and Le Sun},
  338. year={2026},
  339. eprint={2602.22839},
  340. archivePrefix={arXiv},
  341. primaryClass={cs.AI},
  342. url={https://arxiv.org/abs/2602.22839},
  343. }
  344. ```

Ratings and comments

Sign in to rate this skill.

Sign in

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