How long does a typical plan review take?

A thorough review with full source code context and codebase investigation usually takes between 2 to 3 minutes depending on the complexity of the plan.

Why use Gemini to review Claude's implementation plans?

Multi-model validation reduces bias. Using Gemini as a 'second pair of eyes' helps identify specific edge cases or architectural risks that a single model might overlook.

Does this skill require special environment permissions?

Yes, because it uses bash piping and external model calls, it requires the 'dangerouslyDisableSandbox' setting to be enabled in your Claude configuration.

Where are the plan files stored for review?

Claude typically writes its implementation plans to the ~/.claude/plans/ directory. This skill then pipes that content to Gemini for analysis.

Can Gemini modify my codebase during a plan review?

No, the skill is specifically prompted to provide feedback and critique only. It is configured as a read-only advisory layer to ensure safety.

Plan Review via Gemini

Name: Plan Review via Gemini
Author: robbyt

byrobbyt

•

コラボレーションツール

Validates Claude's implementation plans using Gemini for cross-model peer review and risk assessment.

概要

The Plan Review skill integrates Google's Gemini into the Claude Code workflow to provide an objective second opinion on complex implementation plans. By piping Claude's generated plans into Gemini, developers can catch architectural gaps, identify overlooked risks, and ensure edge cases are addressed before execution. This skill acts as a specialized peer review layer, leveraging multi-model intelligence to improve the robustness, safety, and efficiency of automated coding tasks within the Claude environment.

主な機能

Non-destructive feedback loops that prevent unauthorized code changes
Automated completeness checks for edge cases and testing
Deep risk assessment for migrations and breaking changes
Cross-model architectural validation between Claude and Gemini
31 GitHub stars
Context-aware analysis using codebase investigation tools

ユースケース

Validating complex database migration plans before implementation
Ensuring feature development plans are exhaustive and follow best practices
Getting a second perspective on high-risk refactoring strategies

概要

主な機能

Non-destructive feedback loops that prevent unauthorized code changes
Automated completeness checks for edge cases and testing
Deep risk assessment for migrations and breaking changes
Cross-model architectural validation between Claude and Gemini
31 GitHub stars
Context-aware analysis using codebase investigation tools

ユースケース

Validating complex database migration plans before implementation
Ensuring feature development plans are exhaustive and follow best practices
Getting a second perspective on high-risk refactoring strategies