Nothing to hand? Load the — three survey-verb aims, no hypothesis, no preliminary data, headed for a not-competitive outcome — or the , which is close to competitive and where the useful output is what to protect. Both replay a saved review for free.
Paste the section — the prescan is free
No upload, no model: the prescan reads your draft in the browser and checks what can be checked mechanically. Length against the agency's stated limit for that section — NIH Specific Aims at one page, an NIH abstract at thirty lines, an NSF Project Summary at one page, a Research Strategy at twelve, a Project Description at fifteen. The required headings, which for an NSF summary means the literal words Overview, Intellectual Merit and Broader Impacts. The aims: how many, and whether they open on a testable verb or a survey verb. The five NSF broader-impact areas, and whether the activities carry a timeline and an assessment plan. The rigor signals whose absence draws the most common Approach criticism — no alternative if the method fails, no power or sample-size reasoning, no preliminary data, nothing on reproducibility. Citations and figure references. And nine writing checks: undefined acronyms, passive voice, sentences over forty words, hedging, vague quantifiers, reader-dismissing phrasing and average density. Page counts are estimates from word count, because a browser cannot see your font or margins.
The panel rules on it — this is the metered part
A reviewer reads the same section against the agency's published criteria and writes the summary statement: an overall impact score on the NIH one-to-nine convention, applied to all four agencies so two drafts are comparable; a rating and a score for every criterion that agency publishes; a resume of discussion; weaknesses that each quote the verbatim line of your own draft they rest on, with what it costs at review and the concrete fix; and the strengths the revision has to preserve, because a critique that only tears down cannot be worked. Every prescan flag comes back confirmed or explicitly overruled with a reason. Paste the previous summary statement and the panel also rules, criticism by criticism, on whether you actually answered it. Pricing is honest: a worst-case amount is reserved before the run and only what the run uses is charged; the meter next to the button shows both.
Work the plan, then re-check for free
The critique ends in an ordered plan, not a scolding: what to change first, what it costs in effort, and your aims rewritten — statement, rationale, working hypothesis and expected outcome, each opening on a verb a reviewer can test. Take it out in the form that fits what you do next: a revision brief that carries every fix with the line it rests on and the strengths as explicit do-not-touch constraints; a response-to-reviewers scaffold, built in your browser from the findings, with one comment / response / change block per weakness ready for the resubmission introduction; or Markdown, JSON, a findings CSV and print. Rewrite, paste the revised draft back, and the prescan diffs the compliance flags for free — cleared, still open, changed but still firing, newly introduced — before you decide whether a second paid review is worth it. Reviews are saved to your SkillSafe account when you are signed in, and follow you to your other devices.
Derived from the @davila7/research-grants skill. A rehearsal, not an agency review: this critique carries no standing with NSF, NIH, DOE or DARPA, and no tool can tell you whether a proposal will be funded.