深智智预习生成

基于教材PDF或图片OCR识别，AI生成对应学科年级的互动SVG幻灯片与预习小测验课件。

Security Scan

The skill's code and instructions are consistent with a CLI that uploads local PDFs/images to a remote deepaistudy service to produce OCR'd text and interactive slides; nothing in the package appears to perform unrelated or hidden actions, but it requires trusting the remote server and stores credentials locally.

Purpose & Capability

Name/description promise (OCR a textbook and produce interactive SVG slides + quizzes) matches the code: the CLI implements APIs to upload images/PDFs and request generation from a deepaistudy server. The code uses requests to call /api/* endpoints and handles OCR/results via the server. No unexpected external services or unrelated credentials are requested.

Instruction Scope

SKILL.md instructs configuring a server, username and password and uploading local files; the code implements those flows. Minor inconsistency: SKILL.md asks the user to be logged into the website in a browser, but the CLI performs its own API login (/api/mobile/login) using stored credentials — browser login is not required by the code. The skill reads local image/PDF files and transmits them to the configured server (expected for this purpose).

Install Mechanism

No install spec is present in the registry (instruction-only), but the package includes standard Python files and a setup.py that depends only on the widely used requests library. There are no remote downloads or extracted archives referenced in the package. Installing/running the CLI will execute code that performs network I/O (expected).

Credentials

The skill requires a server address plus a username and password (set via the CLI and persisted under ~/.config/deepaistudy-prep/config.ini). These credentials are proportionate to the stated purpose (authenticating to the deepaistudy service), but storing plaintext credentials in a local config file and transmitting them to a remote server are security-sensitive actions — you should only provide credentials for a service you trust and consider using a dedicated/test account or token if possible.

Persistence & Privilege

The skill does not request elevated system privileges, does not set always:true, and only writes its own config under the user's home (~/.config/deepaistudy-prep). It does not modify other skills or system-wide agent settings.

Assessment

This package appears to do what it says: it reads local PDFs/images and uploads them to a configured deepaistudy server which performs OCR and AI generation. Before installing or running it: (1) verify and trust the server URL you configure — do not point it to an untrusted host; (2) prefer creating a dedicated/test account rather than using primary credentials; (3) be aware credentials are stored in ~/.config/deepaistudy-prep/config.ini in plain form — check file permissions or avoid long-term storage; (4) do not upload sensitive documents (student data, proprietary materials) unless the remote service's privacy/security are acceptable; (5) if you need higher assurance, review the remainder of deepaistudy_prep.py (the provided file was truncated) to confirm there are no unexpected network endpoints or telemetry, and consider running the CLI in a sandboxed environment. My confidence is medium because the package source is unknown and storing/transmitting passwords is a practical risk, though the implementation is coherent with the stated purpose.

deepaistudy-prep Skill

深智智智能学习系统预习课件生成 Skill。通过 AI 分析教材 PDF 或图片，生成互动预习课件（SVG幻灯片 + 小测验）。

工作原理

用户图片/PDF → [Qwen VL Max OCR] → 课文原文 → [Gemini SVG生成] → 互动幻灯片 + 小测

前置要求

1. 服务器运行中：本地开发服务器（cd ~/study_ai && ./start_dev.sh）或远程 www.deepaistudy.com
2. 用户已登录：浏览器中已登录深智智网站
3. AI 余额充足：余额不足会导致生成失败

使用方式

方式 1：AI 分析 PDF（推荐）

分析 PDF 并自动识别目录、学科、年级：

deepaistudy-prep analyze /path/to/textbook.pdf --ai

方式 2：上传图片生成预习

deepaistudy-prep upload /path/to/images/*.jpg \
  --subject 数学 \
  --grade "小学三年级" \
  --topic "第一章 加减法" \
  --difficulty medium

方式 3：批量从 PDF 生成（整本书）

deepaistudy-prep batch /path/to/textbook.pdf \
  --subject 数学 \
  --grade "小学三年级" \
  --auto-split

配置

首次使用需要设置服务器地址和认证：

deepaistudy-prep config set server https://www.deepaistudy.com
deepaistudy-prep config set username [e-mail]
deepaistudy-prep config set password your_password

查看当前配置：

deepaistudy-prep config list

常用命令

命令	说明
analyze <file> --ai	AI 自动分析 PDF
upload <path> --subject X --grade Y	上传图片生成预习
batch <file> --auto-split	批量从 PDF 生成
list	查看预习列表
status <prep_id>	查看生成状态
result <prep_id>	获取生成结果
config set <key> <value>	设置配置项
config list	查看当前配置

返回内容

生成完成后返回：

• ocr_text：课文原文（OCR 识别）
• animation_svg：SVG 格式互动幻灯片
• slide_images：渲染后的 PNG 图片
• interactive_preview.quiz：预习小测验题目
• knowledge_points：知识点列表

状态说明

状态	说明
queued	排队中
processing	生成中
completed	完成
failed	失败

注意事项

• 生成大约需要 1-3 分钟（OCR + AI 生成）
• 图片会缩放到最大 2048px
• 支持学科：语文、数学、英语、物理、化学等
• 小测答对才算预习完成