review-recommendations

Review technical recommendations by spawning a general-purpose subagent to challenge assumptions and filter out generic advice, ensuring suggestions are context-specific before deployment. This validates relevance across configuration changes, performance optimizations, and architecture decisions.

A structured verdict process for engineering recommendations evaluates suggestions through problem, context, assumption, and confidence checks to decide whether to keep, modify, or drop the advice before it reaches users.

Filter generic advice from performance optimization suggestions by using a subagent to verify that recommendations match your specific context, discarding or modifying any guidance that fails an assumption check against your environment.

Yes, you can use an automated general-purpose subagent to validate architecture decisions by applying problem, context, and confidence checks to ensure the recommendations are relevant and include proper guardrails before deployment.

You should not use automated recommendation review when configuration changes require immediate execution without a validation layer, as the subagent spawning and structured verdict process adds latency to the decision-making workflow.

The best way to ensure context-specific engineering recommendations is to apply an automated review subagent that performs assumption and context checks, dropping any generic guidance that does not directly apply to the target environment.