AI Dictionary › Grace

Rubric / Evaluation rubric

Rubric / Rubrica di valutazione

In Grace, the evaluation rubric is the structured system of criteria the AI judge uses to assign a score to the user's prompt. It is not a holistic or subjective evaluation: the rubric breaks down prompt quality into 7 measurable dimensions, each with a specific weight on the total score.

Definition

The 7 dimensions are always the same: Role Assignment, Context, Specific Objective, Constraints, Output Format, Audience and Token Efficiency. Their WEIGHT, however, is not fixed: each scenario has a rubric calibrated to its own professional context, so the weights change from scenario to scenario. In the first scenario of the catalog, for example, they are Role Assignment 10%, Context 20%, Specific Objective 15%, Constraints 15%, Output Format 20%, Audience 15%, Token Efficiency 5%; in a scenario where the audience matters more than the format, the balance shifts. For scenarios without a scenario-specific rubric, and for the free trial on the homepage, Grace uses a generic set of weights, the same for everyone. For each dimension, the AI judge assigns a score from 0 to 100, then computes the weighted average using that scenario's weights to get the final score.

The rubric was designed to be universal in structure, the 7 dimensions apply to any type of scenario, professional sector and AI model, and to give actionable feedback: it doesn't just tell you "you did well or poorly," it points to exactly which dimension is costing you points and why. This is what makes learning on Grace concrete, measurable and progressive.

How it's used in Grace

The rubric is the core of Grace's evaluation system. After each scenario you see not only your total score but your score on each of the 7 dimensions. This lets you pinpoint exactly where to improve: if your Context is high (85) but your Constraints are low (40), you know what to focus on in your next session.

Related terms

• Score
• Scenario
• Role Assignment
• Output Format
• Audience (in prompting)

More in Grace

• XP (Experience Points)
• Badge
• Scenario
• Score
• Percentile
• Grace Certification
• Level (Start, Basics, Method, Advanced, Summit)
• Streak

Put it into practice

• Real scenario: apply this concept with an instant AI verdict.
• Daily gym: short drills to train dictionary concepts.
• Playoff game: challenge other players, elimination-style.

📱 Download the Android app (beta) iOS coming soon

Say what you mean. Get what you need.

Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.