review

Audit experimental results by validating metric computations and conclusion correctness.

Updated Apr 29, 2026
One-click install
npx skills add https://github.com/smanist/a-exp --skill review-smanist
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/smanist/a-exp/tree/main/.agents/skills/review
Command: npx skills add https://github.com/smanist/a-exp --skill review-smanist

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Validates experimental results by auditing metric computations and the validity of conclusions.

Core Features & Use Cases

  • Metrics validation: check metric definitions, calculations, and their applicability to the experimental setup.
  • Findings validation: verify that written conclusions are supported by the metrics and methodology.
  • Full-pipeline execution: run metrics first, then findings, or execute both in sequence with a single command.

Quick Start

Invoke the review workflow with a metric pass, a findings pass, or both by invoking /review.

Frequently Asked Questions about review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I audit experimental results to ensure metric calculations are correct?

Audit experimental results by running a metrics validation pass that checks metric definitions, calculations, and applicability to the experimental setup. This verifies metric accuracy before conclusions are drawn.

What is findings validation in experiment quality assurance?

Findings validation verifies that written conclusions are fully supported by the computed metrics and methodology. It enforces structured checks like constraint extraction, attribution tests, and cross-session citation verification.

How do I run full-pipeline validation for both metrics and findings?

Run full-pipeline validation by invoking the review workflow without specifying a mode. This executes metrics validation first, then findings validation sequentially in a single automated command.

Can I validate reproducibility across multiple experiment sessions?

Yes, you can validate reproducibility across experiment sessions using cross-session citation verification. This structured check ensures findings remain consistent and correctly attributed across different experiment runs.

What structured checks are applied during experiment validation?

Structured checks applied during experiment validation include constraint extraction, degeneracy tests, attribution tests, and cross-session citation verification. These enforce metric and finding correctness.

Does experiment validation work with existing experiment tooling?

Yes, experiment validation integrates with existing experiment tooling to support both metrics and findings modes. This allows you to audit results without replacing your current experimental workflow.