https://github.com/466852675/yh-ai-ppt
基于混合 AI 引擎的智能演示文稿生成平台。支持风格参考图复刻、全局视觉配置与智能大纲生成,让小白也能零门槛产出高颜值、专业级的 PPT。
https://github.com/466852675/yh-ai-ppt
genai google-gemini low-code multi-agent openai ppt-generator presentation-maker productivity-tools react slide-deck tailwindcss typescript visual-design vite
Last synced: about 1 month ago
JSON representation
基于混合 AI 引擎的智能演示文稿生成平台。支持风格参考图复刻、全局视觉配置与智能大纲生成,让小白也能零门槛产出高颜值、专业级的 PPT。
- Host: GitHub
- URL: https://github.com/466852675/yh-ai-ppt
- Owner: 466852675
- License: agpl-3.0
- Created: 2026-01-11T23:12:55.000Z (5 months ago)
- Default Branch: main
- Last Pushed: 2026-05-07T18:18:22.000Z (about 1 month ago)
- Last Synced: 2026-05-07T18:40:00.095Z (about 1 month ago)
- Topics: genai, google-gemini, low-code, multi-agent, openai, ppt-generator, presentation-maker, productivity-tools, react, slide-deck, tailwindcss, typescript, visual-design, vite
- Language: TypeScript
- Homepage:
- Size: 327 MB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
- License: LICENSE
- Agents: AGENTS.md
Awesome Lists containing this project
README
**[English](./README.md) | [中文](./README.zh-CN.md)**
# YH-AI PPT - AI-Powered Presentation Design Platform
> Multi-model AI engine with intelligent routing — from a single sentence to a polished presentation in minutes.
[](https://vitejs.dev/) [](#) [](#) [](#) [](#) [](./LICENSE)
[](https://github.com/466852675/YH-AI-PPT/stargazers) [](https://github.com/466852675/YH-AI-PPT/forks) [](https://github.com/466852675/YH-AI-PPT/issues) [](https://github.com/466852675/YH-AI-PPT/releases)
---
## Highlights
- **AI Productivity Pipeline** — One-sentence topic → structured outline → content expansion → AI-generated slides → multi-format export (PDF / image pack / PPTX)
- **Multi-Model Router** — Adapter-based architecture supporting Gemini / OpenAI / Zhipu / DeepSeek / Volcengine / SiliconFlow / ModelScope / local Ollama. Auto-routes by task type (text / image / vision). Hot-swap without restart.
- **Agent Mode** — Natural language conversation drives the entire PPT creation workflow. 9 AI tools (Function Calling), guided/auto dual execution modes, SSE + WebSocket real-time progress.
- **Design Asset Library** — Template gallery, AI-powered visual style extraction from reference images, personal style vault. 4-layer Prompt synthesis engine for consistent visual quality.
- **MinerU Document Parsing** — Industrial-grade PDF/Word structural extraction for content ingestion.
- **SaaS-Ready Infrastructure** — Payment (Alipay/WeChat), credits, VIP membership, refund risk control, messaging center, RBAC with 6 user roles. Configure merchant keys to go live.
---
## Supported AI Providers
[](#) [](#) [](#) [](#) [](#) [](#) [](#) [](#)
---
## Feature Tour
### Dashboard & Project Management
Card-based project kanban with real-time status tracking, priority pinning, quick-preview carousel, and archive management with asset reuse.

| |
|:---:|
|  |
### AI Productivity Pipeline
End-to-end automation: topic → outline → content → images → export (PDF / image pack / PPTX). Supports version rollback and asset reuse.

| | |
|:---:|:---:|
|  |  |
| | | |
|:---:|:---:|:---:|
|  |  |  |
| | |
|:---:|:---:|
|  |  |
### Design Asset Library
Template gallery, AI-powered visual style extraction, rule injection, and personal style vault.

| | |
|:---:|:---:|
|  |  |
### AI Model Router
Configure and hot-swap between 7+ AI providers. Auto-route by task type.

### Admin Panel
Business dashboard, order management, RBAC permissions, credits, AI engine config, and system settings.

| | |
|:---:|:---:|
|  |  |
### Agent Mode
Conversational AI drives the entire PPT creation workflow. Say it, and it's done.

| | |
|:---:|:---:|
|  |  |
---
## Dual Mode
| | Personal | Enterprise |
|---|---|---|
| **Goal** | Create AI-powered presentations for personal use | Run a commercial SaaS platform |
| **Required config** | 5 items (AI Key + JWT + DB) | Full config + merchant keys + SSL |
| **Features** | Outline → content → images → export | All features + admin panel + payment + VIP |
| **Can skip** | Admin panel, payment, VIP, credits | Nothing |
### Minimal Setup (Personal, 5 items)
```env
PORT=1111
DATABASE_URL="file:./dev.db"
JWT_SECRET="your-secret-key-at-least-32-characters"
AI_PROVIDER="Gemini" # or OpenAI, Volcengine, Zhipu, etc.
GEMINI_API_KEY="your-api-key"
```
---
## Quick Start
### Prerequisites
- **Node.js** v18+ (v22+ recommended)
- An AI provider API key (e.g. [Gemini](https://aistudio.google.com/))
Enterprise Database (optional)
SQLite is used by default and requires no setup. For production / multi-user deployments, switch to:
| Database | Recommended For | DATABASE_URL Example |
|---|---|---|
| **PostgreSQL** | Production, high concurrency | `postgresql://user:pass@localhost:5432/yhai_ppt` |
| **MySQL** | Existing infrastructure | `mysql://user:pass@localhost:3306/yhai_ppt` |
| **SQLite** (default) | Personal / development | `file:./dev.db` |
To switch: change `provider` in `server/prisma/schema.prisma` from `"sqlite"` to `"postgresql"` or `"mysql"`, update `DATABASE_URL`, then run `npx prisma db push`.
### Install & Run
```bash
# 1. Install dependencies
npm install && cd server && npm install
# 2. Configure environment
cp server/.env.example server/.env
# Edit server/.env — fill in AI_PROVIDER and your API key
# 3. Initialize database
cd server && npx prisma db push
# 4. Start development servers
# Option A: Windows one-click
一键启动.bat
# Option B: Separate terminals
npm run dev # Frontend → localhost:1000
cd server && npm run dev # Backend → localhost:1111
```
Default admin account: `admin@local` / `admin12345678` (change in production).
---
## Tech Stack
| Layer | Technologies |
|---|---|
| **Frontend** | React 19.2 · Vite 6.2 · Tailwind CSS 4.1 · TanStack Query 5.9 · Framer Motion 12 · WebSocket |
| **Backend** | Express 5.2 · Prisma 6.19 · SQLite / PostgreSQL / MySQL · Winston · Zod 4.3 |
| **AI** | Google GenAI SDK · MinerU · OpenAI Function Calling · SSE streaming |
| **Commerce** | Alipay SDK · WeChat Pay v3 · Nodemailer · express-rate-limit · JWT |
| **Testing** | Vitest · Playwright · Bun Test |
---
## Architecture
```mermaid
graph TD
UI[React Frontend] --> API[Express API Server]
API --> Router{Multi-Model Router}
Router -- text/logic --> Text[Text Models]
Router -- image gen --> Image[Image Models]
Router -- vision --> Vision[Vision Models]
API --> Agent[Agent Service]
Agent -->|Function Calling| Tools[9 AI Tools]
UI <-->|SSE + WebSocket| Agent
```
---
## Roadmap
- [ ] Native editable PPTX export (preserving text layers and vector graphics)
- [ ] Smart layout engine with automatic text-image alignment
- [ ] PowerPoint animation export
- [ ] Real-time multi-user collaboration
- [x] Agent conversational generation mode
- [x] Commercial SaaS infrastructure (payment, credits, VIP, refund)
- [x] Enterprise RBAC permissions and audit logging
- [x] Growth tools (invite rewards, check-in, CRM)
---
## Star History
[](https://star-history.com/#466852675/YH-AI-PPT&Date)
## Contributors
[](https://github.com/466852675/YH-AI-PPT/graphs/contributors)
---
## License
[GNU AGPL-3.0-or-later](./LICENSE)
If you use this project in a public SaaS product, please attribute **"Based on YH-AI PPT"** with a link to the original repository.
---
*YH-AI PPT: Let every presentation resonate.*