Thank you for sharing the Anki deck. I also enjoy using Anki to record insightful ideas from articles I read. However, my programming skills are limited, so I haven’t found a way to batch-generate Anki cards from articles. As a native Chinese speaker currently reading more English content, my current workflow involves using iOS Shortcuts to generate a Markdown file from an article, then sending it to ChatGPT for a Chinese summary. If necessary, I’ll ask ChatGPT to generate a CSV document for importing into Anki.
My prompt ensures the generated cards support both Chinese and English while preserving the original text for easy reference. Here’s the prompt （translated in English）:
```
You are a learning science coach and Anki flashcard design expert. Based on the article content I provided above, please generate high-quality Anki cards according to the following requirements:
------
Card Creation Principles
1. Anki Best Practices for Card Creation:
◦ Generate a separate card for each fact or idea (minimum information principle).
◦ If the original text includes examples, sources, authorities, analogies, metaphors, or contextual details, create additional Q&A cards to ensure no key information is omitted.
◦ Core ideas and supporting details should complement each other; vary question formats, prioritize active recall, and avoid recognition-based or multiple-choice questions.
◦ Provide concise annotations or context for technical terms or ambiguous content.
◦ If a single sentence contains multiple knowledge points, split them into separate cards appropriately—avoid fragmentation or redundancy.
◦ Keep language concise and structure clear.
2. Each card must include an excerpt from the original text for traceability (limit to 200 words).
------
Output Format
Strictly output in pipe-separated CSV format (no headers). Each card occupies one row, with the following fields:
• Field 1: Question (bilingual Chinese & English, separated by <br>)
• Field 2: Answer (bilingual Chinese & English, separated by <br>)
• Field 3: Article URL with text fragment (format: ArticleURL#:~:text=URL-encoded Original Quote—ensure the excerpt is encoded using Python’s urllib.parse.quote() before insertion).
• Field 4: Article title
• Field 5: Author (leave blank if unavailable)
• Field 6: Original excerpt (Original Quote, ≤200 words)
------
Generation Method
1. Display all card content line by line, showing only Field 1 and Field 2 (in table format for easy review).
2. Simultaneously provide the full card data as a pipe-separated CSV file for download.
3. If a single original excerpt can be split into multiple cards, output all of them—one per row.
------
Language Requirements
• All Q&A must be bilingual (Chinese & English)—separated by <br>, with Chinese first and English supplementary.
• If the original text is in Chinese, output only Chinese (no English needed).
• Every card must retain the original excerpt (Field 6).
------
Technical Requirements
• When generating Field 3, ensure the Original Quote is URL-encoded using Python’s urllib.parse.quote().
• Directly output the final encoded result—no need to explain the encoding process.
Thank you for sharing the Anki deck. I also enjoy using Anki to record insightful ideas from articles I read. However, my programming skills are limited, so I haven’t found a way to batch-generate Anki cards from articles. As a native Chinese speaker currently reading more English content, my current workflow involves using iOS Shortcuts to generate a Markdown file from an article, then sending it to ChatGPT for a Chinese summary. If necessary, I’ll ask ChatGPT to generate a CSV document for importing into Anki.
My prompt ensures the generated cards support both Chinese and English while preserving the original text for easy reference. Here’s the prompt （translated in English）:
```
You are a learning science coach and Anki flashcard design expert. Based on the article content I provided above, please generate high-quality Anki cards according to the following requirements:
------
Card Creation Principles
1. Anki Best Practices for Card Creation:
◦ Generate a separate card for each fact or idea (minimum information principle).
◦ If the original text includes examples, sources, authorities, analogies, metaphors, or contextual details, create additional Q&A cards to ensure no key information is omitted.
◦ Core ideas and supporting details should complement each other; vary question formats, prioritize active recall, and avoid recognition-based or multiple-choice questions.
◦ Provide concise annotations or context for technical terms or ambiguous content.
◦ If a single sentence contains multiple knowledge points, split them into separate cards appropriately—avoid fragmentation or redundancy.
◦ Keep language concise and structure clear.
2. Each card must include an excerpt from the original text for traceability (limit to 200 words).
------
Output Format
Strictly output in pipe-separated CSV format (no headers). Each card occupies one row, with the following fields:
• Field 1: Question (bilingual Chinese & English, separated by <br>)
• Field 2: Answer (bilingual Chinese & English, separated by <br>)
• Field 3: Article URL with text fragment (format: ArticleURL#:~:text=URL-encoded Original Quote—ensure the excerpt is encoded using Python’s urllib.parse.quote() before insertion).
• Field 4: Article title
• Field 5: Author (leave blank if unavailable)
• Field 6: Original excerpt (Original Quote, ≤200 words)
------
Generation Method
1. Display all card content line by line, showing only Field 1 and Field 2 (in table format for easy review).
2. Simultaneously provide the full card data as a pipe-separated CSV file for download.
3. If a single original excerpt can be split into multiple cards, output all of them—one per row.
------
Language Requirements
• All Q&A must be bilingual (Chinese & English)—separated by <br>, with Chinese first and English supplementary.
• If the original text is in Chinese, output only Chinese (no English needed).
• Every card must retain the original excerpt (Field 6).
------
Technical Requirements
• When generating Field 3, ensure the Original Quote is URL-encoded using Python’s urllib.parse.quote().
• Directly output the final encoded result—no need to explain the encoding process.
------
```