Prompt coach

Grade a prompt before you send it. Your pasted text stays in this browser. It is not sent to a model or stored.

Your prompt and its grade

Your text is graded against the prompt rubric bundled with this page.

Paste a prompt or short conversation, then select “Grade this prompt” to get a score out of 100 and the first change to make. Revise it and grade it again to compare the results.

See how the score is measured and what to do first.

Bundled synthetic example, grading now

This result uses a bundled synthetic example written for this page, not your prompt or anyone else’s real prompt.

Compare a revision. A second grade shows both scores and what changed.

Label turns “User:” and “Assistant:” to grade an exchange, or paste one prompt on its own.

Model fit is one of the three scores. Pick the tier by what the model is at its provider: Premium or frontier is the most capable model a provider sells, Standard is a provider’s general-purpose default, and Economy or small is a provider’s small or cheap model. Left as Not specified, this grade makes no routing recommendation rather than assuming a tier.

What the coach reads and keeps

The coach reads only the text you paste and the optional model tier. It does not access your accounts, files, or customer data.

See the bundled example text and the counts read from it.

See bundled examples of a graded prompt, a prompt that needs changes, and text the coach cannot grade, none taken from text you paste.

Loading results from the bundled examples.

Also on this site · Runs on your own work

Grading more than one prompt?

This page grades one text. Personal AI history runs the same rubric across your assistant export — weeks of your prompts — and names the single habit worth changing first. It is read in your browser too.

Asking where your AI spend goes instead? AI FinOps analyzes a provider export to find AI-spend savings opportunities. Your files stay in that browser tab.

AI FinOps reads a provider export from OpenAI, Anthropic, Azure OpenAI, Google Vertex AI, or AWS Bedrock, as CSV, TSV, JSON, or JSONL. No export to hand? The bundled synthetic example opens on a finished result, so you can read the whole analysis before you find a file.

Open AI FinOps