gemini-cli

Run sequential Gemini reviews across multiple targets with rate limit awareness, model fallback, and an aggregate summary table. Use when reviewing multiple directories, modules, or files in batch.

Use this agent when the user wants to delegate a large context review to Gemini CLI. Trigger phrases: "review with gemini", "gemini review this", "use gemini to analyze", "second opinion from gemini", "gemini code review", "large file review". IMPORTANT: Only spawn ONE instance of this agent per review request. Do not spawn multiple gemini-review agents in parallel — this wastes API quota and creates race conditions. <example> Context: User wants to review a large codebase module user: "use gemini to review the src/api directory for security issues" assistant: "I'll use the gemini-review agent to run a comprehensive security review via Gemini CLI." <commentary> Large context review request targeting Gemini CLI. Delegate to the review agent which orchestrates the headless CLI call. </commentary> </example> <example> Context: User wants a second AI opinion on their code user: "get a second opinion from gemini on this auth implementation" assistant: "I'll use the gemini-review agent to get Gemini's analysis of the authentication code." <commentary> Second opinion request. The review agent will gather the relevant files and send them to Gemini for independent analysis. </commentary> </example> <example> Context: User wants to analyze large log files user: "have gemini analyze the last 50k lines of the app log for error patterns" assistant: "I'll use the gemini-review agent to process the log file through Gemini's large context window." <commentary> Large context analysis that benefits from Gemini's 2M token window. Delegate to review agent. </commentary> </example>

Use this agent when the user says "gemini not working", "fix gemini cli", "gemini error", "nano-banana not working", "gemini auth error", "gemini cli issue", or has problems with Gemini CLI installation, authentication, or extensions. <example> Context: User can't run gemini commands user: "gemini cli not working" assistant: "I'll use the gemini-troubleshoot agent to diagnose the issue." <commentary> Troubleshooting request detected. Delegate to debugging agent. </commentary> </example> <example> Context: User has authentication issues user: "gemini says authentication failed" assistant: "I'll use the gemini-troubleshoot agent to investigate the auth configuration." <commentary> Auth error reported. Delegate to troubleshoot agent for systematic diagnosis. </commentary> </example>

This skill should be used when the user asks to "use gemini", "gemini cli", "configure gemini", "set up gemini", "gemini review", "gemini images", or mentions general Gemini CLI usage. For specific topics, focused skills may be more appropriate.

This skill should be used when the user asks to "generate image with gemini", "create icon", "nano-banana", "gemini image", "generate pattern", "create diagram with gemini", "app icon", "generate favicon", or wants to use the nano-banana Gemini CLI extension for image generation, icon creation, pattern design, or visual content.

This skill should be used when the user asks to "review with gemini", "gemini code review", "analyze large file with gemini", "gemini review codebase", "large context review", "second opinion from gemini", or wants to use Gemini CLI for reviewing code, documents, logs, or other large context items.

1 hook across 1 event