scrutinize

Verify plans, PRs, and code changes against stated intent and execution paths.

1|Updated May 24, 2026
One-click install
npx skills add https://github.com/tinner-deinno/innova-skills-lib --skill scrutinize-tinner-deinno
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: scrutinize
Source: https://github.com/tinner-deinno/innova-skills-lib/tree/main/core/scrutinize
Command: npx skills add https://github.com/tinner-deinno/innova-skills-lib --skill scrutinize-tinner-deinno

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It prevents costly approvals by forcing an outsider, end-to-end review that checks whether a plan/PR/design actually achieves its claimed outcomes.

Core Features & Use Cases

  • Outsider-first intent check: states the goal in one sentence and immediately questions whether a simpler or more elegant approach could achieve the same result.
  • Trace the real code path: follows the call graph through unchanged code to verify behavior, not just changes shown in a diff.
  • Evidence-based verification: separates what the artifact claims from what is actually produced, including edge cases, error paths, and silent side effects.
  • Actionable findings: produces concise findings ordered by severity, each with what to change, why it matters, and the evidence behind it.

Quick Start

Trigger the scrutinize skill when you want a second-opinion audit of a proposed code change by sending the command: "/scrutinize review this proposed change and verify its claims end-to-end."

Frequently Asked Questions about scrutinize

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

FAQPage Schema
How do I perform an end-to-end verification of a pull request against its stated intent?

End-to-end verification of a pull request requires an outsider-perspective audit that checks whether the code changes actually achieve their claimed outcomes by tracing the real execution path. This process evaluates intent, follows the call graph, and verifies behavior.

What is PR auditing and how does it identify silent side effects in a diff?

PR auditing is an evidence-based review that separates what a code artifact claims from what it actually produces. It identifies silent side effects by tracing the call graph through unchanged code, verifying behavior beyond surface-level edits shown in the diff.

Can I use a design sanity-check to find simpler alternatives to a proposed software behavior?

Yes, a design sanity-check states the goal in one sentence and immediately questions whether a simpler or more elegant approach could achieve the same result. It forces an outsider-first intent check to prevent costly approvals of proposed behaviors.

How do I conduct a root-cause trace for edge cases and error paths in a code review?

Conducting a root-cause trace during code review involves following execution paths to explicitly handle edge cases and error paths. It requires evidence-cited findings that detail what to change, why it matters, and the proof behind it.

What is the best way to structure risk assessment findings for a proposed code change?

The best way to structure risk assessment findings is to order them concisely by severity. Each finding must include actionable steps on what to change, why it matters, and the evidence verified during the end-to-end audit.