Three Must‑Try Open‑Source AI Tools for Scraping, PPT Creation, and Video Generation

In the era of abundant AI tools, several open‑source projects on GitHub address practical pain points. This article reviews three recently popular AI tools covering data collection, PPT generation, and video creation, all free and open‑source.

1. Xiaohongshu Full‑Stack Operation Tool: Spider_XHS

Spider_XHS is a high‑star (4000+) crawler that supports comprehensive data acquisition from the Xiaohongshu platform, including note details, user profiles, and media, with export to Excel or local folders.

Key Features

Multi‑dimensional data collection : captures titles, descriptions, tags, likes, comments, shares, user info, and watermark‑free images/video; results exportable.

Automated content publishing : integrates Xiaohongshu creator API, supports QR‑code or SMS login, uploads image sets or videos, and shows published items and messages.

Targeted scraping : can fetch all notes of a user, homepage recommendations, or channel data for various operation scenarios.

Quick‑Start Guide

Environment preparation : install Python 3.7+ and Node.js 18+.

Project deployment :

# Clone the repository
git clone https://github.com/cv-cat/Spider_XHS.git
cd Spider_XHS
# Install dependencies
pip install -r requirements.txt
npm install

Configure Cookie : log in to the web version, open developer tools, copy the Cookie from the Network panel, and paste it into a .env file at the project root.

Run the tool : modify the call logic in main.py if needed, then execute python main.py. Collected data are saved automatically.

Open‑source repository: https://github.com/cv-cat/Spider_XHS

2. AI‑Powered PPT Generation: PPTAgent

PPTAgent, released by the Institute of Computing Technology, Chinese Academy of Sciences, creates professional slides from a single textual command or by ingesting long documents such as papers or reports.

Key Features

Multi‑scenario generation : generate slides from short prompts (e.g., “introduce Xiaomi SU7 design and price”) or from uploaded long texts using Retrieval‑Augmented Generation to extract key points.

Intelligent creation workflow : a multi‑agent architecture mimics human PPT design, handling topic decomposition, content organization, and layout automatically, with styles adaptable to business or academic contexts.

High‑compatibility output : exports native .pptx files that can be edited freely in PowerPoint without formatting issues.

Typical Use Cases

Business reporting: upload annual financial statements to generate data‑visualization slides.

Academic presentations: import a paper to produce background, methodology, and conclusion slides.

Classroom teaching: input a topic such as “decode the legislative process’s impact on international relations” to create teaching materials.

Open‑source repository: https://github.com/icip-cas/PPTAgent

3. Code‑Driven Video Generation: Code2Video

Code2Video, from a National University of Singapore team, uses the Manim animation engine to produce high‑quality instructional videos directly from code, with AI agents generating the necessary Manim scripts.

Key Features

Unique generation logic : leverages Manim to control each frame via code, yielding logically rigorous and detail‑accurate videos.

AI‑assisted scripting : an AI agent creates Manim code from natural‑language requests, lowering the barrier for non‑programmers.

HD watermark‑free output : renders clear, watermark‑free videos suitable for tutorials, demos, and knowledge sharing.

The project quickly earned 1.5K+ stars and was accepted by the NeurIPS 2025 Workshop (DL4C), indicating academic recognition.

Open‑source repository: https://github.com/showlab/Code2Video

All three projects are easy to deploy, highly practical, and useful for developers and professionals seeking to boost productivity in content operation, office efficiency, and knowledge dissemination.