jlpt-n5-grammer-text-integration-question-creator

name: jlpt-n5-grammer-text-integration-question-creator description: Generate JLPT N5 text-level grammar (Mondai 3) questions in JSON format. Use this skill when the user asks to create, generate, or produce N5 text-level grammar questions, text integration exercises, or "metin tamamlama" questions. Output is a JSON object with a "textFlow" type, containing "title", "textSegments", and "blanks" array. Trigger on requests like "N5 Mondai 3 sorusu üret", "create N5 text integration questions", "generate text-level grammar", or "JLPT N5 metin bütünleme sorusu oluştur".

JLPT N5 Text Integration Question Creator

Generate N5-level text-level grammar (Mondai 3) questions as JSON.

References

• Output format & field rules: See references/json-schema.md
• N5 grammar points and connectors: See references/n5-grammar-points.md

Workflow

1. Determine the theme for the short text (e.g., daily routine, email, announcement, short story) from user request or context.

2. Read references/json-schema.md for textFlow field specifications.

3. Read references/n5-grammar-points.md when selecting connectors (shikashi, soshite, etc.) or verb forms.

4. Generate the question:

   • Create a title for the text.
   • Write a natural N5-level short text (approx. 100-200 characters).
   • Identify 3 points in the text for blanks (usually connectors, particles, or verb/adjective endings).
   • Split the text into textSegments array. The number of segments will be number of blanks + 1.
     • For each blank:
     • Choose 4 options: 1 correct + 3 plausible N5 distractors.
     • Set blankNumber starting from 1 for each text (1, 2, 3...).
     • Set position (0, 1, 2...).
     • Set id as n5_grammer_text_integration_XXX where XXX is a zero-padded sequential number. Continue from the last existing ID in the file.

5. Append the generated question to the questions array in backend/grammar/data/n5_text_integration.json. Read the file first, then add the new question to the end.

6. Validate: After saving, run the jlpt-n5-grammer-text-integration-question-tester skill to validate the JSON file. If any questions fail, fix them and re-run the tester until all pass.

7. Upload prompt: After all questions pass validation, ask the user whether to upload to Firebase. If yes, run: python3 backend/grammar/scripts/upload_grammar_questions.py n5 --type text-integration

8. Output a summary of the text created and the grammar points tested.

Quality rules

• All vocabulary and kanji must be within N5 scope.
• Text must be a coherent narrative, not disconnected sentences.
• Connectors (but, then, therefore) are primary targets for Mondai 3.
• Ensure the logical flow only allows one correct option for each blank.
• Default: 1 text with 3 blanks per request unless user specifies otherwise.