Evaluation Rubrics

What it does

This skill allows AI agents to generate and apply evaluation rubrics. It can create rubrics based on provided criteria, assess outputs against those rubrics, and provide scores with justifications. The skill is designed to standardize assessment processes and offer more detailed feedback than simple scoring.

When to use it

• Grading student work: Generate a rubric for an essay or project and then use the AI agent to score submissions based on that rubric.
• Code review: Create a rubric outlining coding best practices, then have the agent evaluate code snippets against those standards.
• Content quality assessment: Develop a rubric defining criteria for high-quality content (e.g., clarity, accuracy) and use it to assess articles or reports.
• Evaluating AI model outputs: Define specific performance metrics in a rubric format and utilize the skill to evaluate different model responses objectively.

Key capabilities

• Rubric generation based on provided criteria.
• Assessment of outputs against defined rubrics.
• Scoring with justifications tied to rubric elements.
• Standardized assessment processes.

Example prompts

• "Create a rubric for evaluating the clarity and completeness of technical documentation."
• "Assess this essay [essay text] using the following rubric: [rubric details]."
• "Score these code snippets [code snippets] based on a rubric emphasizing readability and efficiency."

Tips & gotchas

The quality of the generated rubrics and assessments heavily depends on the clarity and specificity of the input criteria. Providing detailed instructions or example outputs will improve results.