Free to start
Grade your prompt. Free to start.
Paste an instruction. Get a rubric score, the top three weaknesses, and a sharper rewrite in 20 seconds. Sign in to grade your first one free.
Paste a prompt, command, skill, agent spec, or rubr_flow block…
The three weaknesses, structural ones first.
The deterministic, structural problems lead — a missing output contract, no success criteria, an unbounded step. Those are checks, not opinions, which is why an inline model will not reliably give them to you.
No output contract — the shape of a correct answer is never stated.
No success criteria — nothing says when the result is done.
Tone is left implicit, so results drift run to run.
Want the eval that proves the rewrite holds, version history, and a proof report? Sign in.
What the free grade gives you.
The free grade is the wedge: enough to see your instruction is weaker than it felt, and exactly where. The loop — proof you can keep and share — lives in the account.
Free to start
A rubric score, the top three weaknesses, and a sharper rewrite. No credit card to begin.
Structural problems first
The deterministic findings lead, because they are the ones an inline "improve this" will not reliably give you.
Then the full loop
Sign in for the eval that proves the rewrite holds, version history, and shareable proof reports.
Any artifact type
Prompt, command, skill, agent spec, workflow, or rubr_flow — the type is detected and graded against the matching rubric.
Answers before you start.
Yes. Create a free account and grade your first instruction — a rubric score, the top three weaknesses, and a sharper rewrite — no credit card. Usage is credit-metered after that so you can see exactly what each run costs.
Know which instructions are ready to run.
Grade your promptFollow the review loop as it ships.
Notes on AI artifact testing, rubr_flow conversion, evals, and proof reports.