gemini-rescue

To get a second opinion on a code diff, this Skill hands off the current diff or plan to Google Gemini for an independent adversarial review. It surfaces alternative approaches and cross-vendor validation when Claude is stuck.

Cross-vendor validation is the process of using a different AI vendor, like Gemini, to audit high-sensitivity changes such as auth or migrations. It provides an independent read to catch errors the primary model might miss.

To review long-context code or plans, this Skill leverages Gemini's 1M-context window for comprehensive audits. It passes the extensive data to Gemini to perform an independent analysis that standard context windows cannot handle.

Yes, you need a registered Gemini MCP server exposing the gemini_ask and gemini_run tools to perform the independent code review. This server acts as the bridge to dispatch the analysis and retrieve actionable recommendations.

Gemini can do both: Mode A provides a one-shot review with structured output, while Mode B enables agentic dispatch where Gemini reads files and proposes actual patches. Choose the mode based on your review depth requirements.

Use an adversarial review approach for high-sensitivity tasks, complex migrations, or when your primary model is stuck. It forces an independent alternative analysis to surface potential flaws and validate strategies across vendors.