ai-code-review-refactor

What problem does it solve?

It helps you prevent broken or risky refactors by reviewing AI-generated or heavily assisted code for hallucinated APIs, hidden assumptions, architecture drift, dependency bloat, missing tests, type unsafety, and unsafe behavioral changes.

Core Features & Use Cases

• Verify real API existence: Checks that imports, method calls, and parameter signatures exist in the declared versions to stop runtime failures from hallucinated code.
• Enforce safe refactor boundaries: Ensures behavioral equivalence and surfaces any silent changes to observable behavior as reviewable, separated work.
• Require evidence-backed completion: Demands an evidence inventory and an Execution Discipline Report so “completion” claims are grounded in validation outputs rather than plausibility.
• Harden security and reliability: Flags swallowed exceptions, silent null returns, insecure paths (auth/permissions/data access), and missing adversarial review where needed.

Quick Start

Ask an AI to review a proposed refactor of AI-assisted code and return a checklist of verified APIs, hidden assumptions, dependency risks, missing tests, and an explicit refactor boundary assessment before any changes are accepted.