cross-modal-review

Validate work products against contracts using a second AI model.

5|1|Updated Apr 15, 2026
One-click install
npx skills add https://github.com/GYF0311/lorekit --skill cross-modal-review-gyf0311
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: cross-modal-review
Source: https://github.com/GYF0311/lorekit/tree/main/brain/skills/cross-modal-review
Command: npx skills add https://github.com/GYF0311/lorekit --skill cross-modal-review-gyf0311

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It prevents low-quality or contract-violating work from being finalized by requiring an independent review from a different model with refusal routing.

Core Features & Use Cases

  • Second-model quality gate: Grades the work against the originating skill’s Contract (not “vibes”) and reports PASS/ISSUES with evidence.
  • Refusal routing: Silently switches to the next model if the first refuses, without exposing refusals to the user.
  • Code-review handoff guidance: Recognizes when to recommend Codex /codex review for diff review and can switch to an adversarial challenge mode for security-sensitive changes.

Quick Start

Ask your AI to perform a cross-modal review of your proposed change for contract compliance before you commit it.

Frequently Asked Questions about cross-modal-review

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

FAQPage Schema
How do I get a second opinion on code changes from a different AI model?

To get a second opinion on code changes, this cross-modal review validates your work against the originating skill's contract and reports PASS or ISSUES with evidence before finalizing.

What is contract-based grading in AI code review?

Contract-based grading in AI code review evaluates work products against the originating skill's predefined contract, providing evidence-based PASS or ISSUES findings instead of subjective feedback.

How do I handle AI model refusals during security validation?

During security validation, refusal-chain routing silently switches to the next available AI model if the first one refuses, ensuring continuous review without exposing refusals to the user.

When should I use a second AI model for code review?

Use a second AI model for code review during security-sensitive edits, stuck iterations, pre-bulk operations, or when applying new skills that affect behavior to prevent finalizing contract-violating work.

How do I run an adversarial challenge for security-sensitive code edits?

For security-sensitive code edits, the cross-modal review can switch to an adversarial challenge mode and recommend dedicated diff review tools like Codex.

Does the AI reviewer automatically apply fixes to my code?

No, the AI reviewer does not automatically apply fixes because it enforces strict user sovereignty, meaning all reviewer findings remain informational until explicitly approved by you.