"Review this paper for chemistry quality" lives in your head as one verb. The careful version is actually six small checks done in order. Write them down once, and the skill runs them every time without you having to decide where to start.
The pattern: 3 to 6 steps, each one verb-led, each one you could ask "did I do this?" If a step starts with "be" or "consider" or "think about" — it's not a move, it's an aspiration. Rewrite it as a verb.
Doesn't tell anyone what a thorough reviewer actually does. Every run lands somewhere different.
Each step is a verb. Each step is checkable. Every run produces the same shape of output.
Review this paper for chemistry quality.
1. Check the synthesis is reproducible: precursors, solvents, conditions, scale, atmosphere all specified. 2. Verify the control experiments actually distinguish the proposed mechanism from the obvious artifact (solvent, background, blank). 3. Audit the characterization (XRD, NMR, MS, XPS) against the claimed structure. Are all peaks accounted for? 4. Cross-check yield / purity claims against the SI data. Do the integrations support the headline numbers? 5. Flag any critical claim that rests on a single replicate. 6. Identify whether the conclusion overreaches the data.
Each step is a verb. Each step is checkable. You could run this skill on any chemistry manuscript and the output would have the same shape — which is what makes the rejection letter you draft from it defensible.
Don't start with the perfect list. Start with "what do I actually do, in order, when I do this well?" Write five lines. Verbs at the front. If any line starts with "consider" or "think about" — rewrite it.