Gemini Peer Review

The gemini-peer-review skill enables AI agents to act as a secondary reviewer, analyzing code or text generated by other models to identify errors, suggest improvements, and verify logic before final implementation. It simulates a collaborative development environment where one agent critiques the work of another to enhance overall output quality.

When to use it

• Before deploying critical code snippets to catch syntax errors or logical flaws early in the workflow.
• When generating documentation or technical reports that require rigorous fact-checking and clarity reviews.
• During iterative design processes where an initial draft needs refinement based on specific architectural constraints.
• To validate complex reasoning chains or mathematical proofs generated by a primary agent.

Key capabilities

• Automated critique of code quality, style, and potential bugs.
• Identification of logical inconsistencies in text-based arguments or plans.
• Suggestion of concrete improvements for clarity, efficiency, or best practices.
• Verification of outputs against predefined standards or requirements.

Example prompts

"Act as a peer reviewer and analyze the following Python function for edge cases and performance bottlenecks." "Review this draft technical specification for clarity and ensure all constraints are addressed before we proceed to implementation." "Simulate a code review session where you critique my recent changes against our team's style guide and security protocols."

Tips & gotchas

Ensure the primary agent providing the work for review has clear context about the project goals, as the peer reviewer relies on this information to provide relevant feedback. Limit the scope of reviews to specific artifacts (e.g., "review only the authentication module") to prevent the AI from losing focus or hallucinating constraints outside its immediate analysis.