wtf is nature-reviewer-skills?

Nature Reviewer Skills is a set of tools built to help researchers stress test a scientific manuscript before it goes to actual journal reviewers. Rather than checking grammar or offering a generic review my paper prompt, it applies discipline specific review checks that look for the kinds of weaknesses that sink papers in high level peer review: claims not backed by enough evidence, incomplete chains of reasoning, weak comparison groups, conclusions that go beyond what the data actually shows, and uncertainty that was never properly measured. The guiding idea is that stronger claims need stronger evidence to back them up. The tool examines a manuscript across several angles at once, including whether the central claim is properly supported, whether the evidence chain from data to conclusion holds together, whether controls and baselines can actually rule out other explanations, whether validation testing is independent and representative, whether uncertainty is properly accounted for, whether the evidence shows real causality or just correlation, and whether the results are being generalized further than the data supports. Running it produces two to four separate reviewer style reports, each written from a different angle, rather than one flat checklist. Every major concern raised is expected to include the specific claim being questioned, the exact evidence it relates to, why the issue matters, how severe it is, a possible alternative explanation, and a concrete suggestion for how to fix it. The suite includes seven subject specific reviewer skills covering remote sensing, atmospheric science, hydrology, climate and ecology, chemistry, engineering, and materials science, plus an additional orchestrator focused specifically on polar and Arctic or Antarctic research that routes claims to the right domain reviewers and applies extra polar specific checks. Written in Python, this project is aimed at researchers preparing papers for selective journals, principal investigators running internal reviews, graduate students learning how rigorous scientific criticism works, and developers building their own AI based review systems.