Open evaluation workspace
Create a judge score.
Paste content or upload text, Markdown, or JSON files. Define what good looks like in your rubric, then create a structured evaluation result.
🔒 Files stay in your browser
Your score will appear here
Add a request, response, and rubric—or load the sample—to create an evaluation.
Overall score
—/10
Criterion scores
Judge reasoning
Demo scores use a deterministic browser evaluator, not an external LLM. Connect the open-source Python template to a judge model for semantic production evaluation.