Use this prompt with ChatGPT, Claude, or any LLM to generate a properly formatted rubric. Then paste the JSON output below.
Prompt Template
Example Output
I need you to create a JSON evaluation rubric for assessing [YOUR TOPIC HERE, e.g., "customer support quality", "code review feedback", "essay writing"].
The rubric should follow this exact JSON structure:
{
"id": "unique_snake_case_id",
"name": "Human Readable Rubric Name",
"description": "A clear description of what this rubric evaluates and when to use it.",
"usage_instructions": "Detailed instructions for evaluators on how to apply this rubric consistently.",
"scale": {
"min": 0,
"max": 5,
"labels": {
"0": "Worst level label",
"1": "Poor level label",
"2": "Below average label",
"3": "Acceptable label",
"4": "Good label",
"5": "Excellent label"
}
},
"dimensions": [
{
"id": "dimension_id",
"name": "Dimension Name",
"description": "What this dimension measures",
"criteria": [
{
"score": 0,
"label": "Short label for score 0",
"description": "Detailed description of what warrants this score",
"examples": [
"Specific example 1",
"Specific example 2"
]
},
// ... repeat for scores 1-5
]
}
// ... add 3-6 dimensions
]
}
Requirements:
1. Include 3-6 evaluation dimensions relevant to [YOUR TOPIC]
2. Each dimension must have criteria for all scores 0-5
3. Include 2-3 concrete examples per criterion
4. Make descriptions specific and actionable
5. Ensure criteria are mutually exclusive (no overlap between score levels)
6. Output ONLY valid JSON, no markdown code blocks
Please generate the rubric now.
{
"id": "email_communication",
"name": "Email Communication Rubric",
"description": "Evaluates professional email communication quality...",
"usage_instructions": "Score each email against all dimensions...",
"scale": {
"min": 0,
"max": 5,
"labels": {
"0": "Unacceptable",
"1": "Poor",
"2": "Needs Improvement",
"3": "Acceptable",
"4": "Good",
"5": "Excellent"
}
},
"dimensions": [
{
"id": "clarity",
"name": "Clarity & Conciseness",
"description": "How clear and concise is the message?",
"criteria": [
{
"score": 5,
"label": "Crystal Clear",
"description": "Message is immediately understandable...",
"examples": ["Well-structured with bullet points..."]
}
// ... more criteria
]
}
// ... more dimensions
]
}