All products

Peer review a data analysis with AI

Agree the review scope, document specific issues and give the analyst a way to answer each one. Close the review only when the material issues are resolved.

Peer review a data analysis with AI when a colleague needs more than general comments on a finished report. The peer-review-template skill helps the two of you agree whether the review covers methods, logic, code, reproducibility or presentation. It then splits feedback by importance.

A change to a chart label may be worthwhile, but it should not obscure a shift in the underlying denominator. If the conclusion depends on the denominator, say where the definition changed, show the evidence and explain what needs checking. The author can respond to that point even if they ultimately defend the original choice.

What it does

  • Check whether the method answers the question the analysis set out to address.
  • Review assumptions and the support for major conclusions.
  • Inspect analytical code or SQL where it falls inside the agreed scope.
  • Distinguish must-fix issues from useful suggestions and optional polish.
  • Track the author’s response and record the outcome for each material point.

Install and use

  1. Install the skill in a compatible agent. Provide the authorised report, notebook, specification or SQL and a review scope agreed with its author. No review service is required.
  2. Ask, “Check the validity of this month-on-month comparison and keep presentation suggestions separate.”
  3. Give the author the findings. Have the reviewer confirm the changes or rationale before signing off.

FAQ

Source

Made by Nimrod Fisher. Original skill: peer-review-template, published on GitHub under the MIT licence.

View on GitHub