Rate a prompt
Paste a prompt and it gets scored on eight named dimensions, with a comment on each and one rewritten version demonstrating the fixes. The rubric reweights for the model you are targeting.
Composing a prompt here needs no account. Seeing the scored result does - it is the only part of Prompt Corpus that costs money to run, which is why it is capped at 20 ratings per person per day. The reasoning behind every dimension is in Evaluating prompts systematically.
Readout
Eight dimensions, scored separately. One composite number would tell you nothing you could act on, so you get the breakdown and a rewritten version demonstrating the fixes.
- Clarity
- Specificity
- Context
- Structure
- Constraints
- Examples
- Role framing
- Task decomposition