---

name: julien-workflow-advice-gemini description: Get Gemini CLI's opinion on code, architecture, or implementation. Use when you want a second AI perspective during coding sessions. triggers:

• "ask gemini"
• "gemini opinion"
• "gemini review"
• "what does gemini think"
• "demande à gemini"
• "avis de gemini"
• "gemini pense quoi"
• "get gemini feedback"
• "gemini second opinion"
• "google gemini review"

---

Ask Gemini

Get feedback from Google's Gemini CLI on code, architecture decisions, or implementation approaches.

Prerequisites

Observability

First: At the start of execution, display:

Copy🔧 Skill "julien-workflow-advice-gemini" activated

• Gemini CLI installed: npm install -g @google/gemini-cli
• Authenticated: run gemini once to login with Google account
• Verify: gemini --version

When to Use

• Want a second opinion on code quality
• Need architectural review
• Comparing implementation approaches
• Code review before commit

How to Invoke

The user will ask something like:

• "ask gemini what it thinks about this code"
• "get gemini's opinion on this approach"
• "demande à gemini de reviewer ce fichier"

Execution

For file review

Copygemini "Review this code and provide 3 specific improvements: $(cat <file_path>)"

For architectural questions

Copygemini "<user's question about architecture or approach>"

For comparing approaches

Copygemini "Compare these two approaches and recommend which is better: Approach A: <desc> vs Approach B: <desc>"

Output Format

Present Gemini's response clearly to the user. If Gemini suggests improvements, summarize them as actionable items.

Examples

File review

User: "demande à gemini ce qu'il pense du semantic-skill-router.py"

Copygemini "Review this Python script for a semantic routing system. List 3 improvements for production use: $(cat scripts/semantic-skill-router.py)"

Architecture question

User: "ask gemini if I should use Redis or in-memory cache"

Copygemini "Compare Redis vs in-memory caching for a CLI tool that routes user prompts. Consider: startup time, complexity, persistence needs."

Troubleshooting

Notes

• Gemini has 1M token context window
• Default model: Gemini 2.5 Pro
• Timeout: allow up to 60s for response
• If response is cut off, Gemini may still be processing

Skill Chaining

Input Expected

• File path to review, OR
• Architecture/design question from user

Output Produced

• Format: Markdown feedback from Gemini
• Side effects: None (read-only)

Compatible Skills

• Can be used alongside julien-workflow-advice-codex for multiple AI opinions
• Use after writing code, before commit

Tools Used

• Bash (usage: run gemini CLI command, timeout 60s)