reflexion-reflect

To critically evaluate and improve AI-generated outputs, apply a structured self-reflection framework that checks for completeness, quality, correctness, and dependencies. This process identifies specific flaws and proposes targeted solutions for iterative enhancement.

The best way to self-review generated code is to apply criteria for code-specific reflection, checking architectural alignment, correctness, and efficiency. This ensures the final product adheres to established best practices through a rigorous refinement process.

Complexity triage for AI output reflection adjusts the evaluation depth based on the task's difficulty, categorizing it as Quick, Standard, or Deep. This ensures the self-assessment process matches the rigor required by the specific problem.

Yes, you can fact-check claims and verify dependencies by applying detailed checks for correctness and completeness during the self-reflection process. This evaluates previous responses against defined criteria to ensure accuracy and architectural alignment.

You do not need a specific external framework; iterative improvement for software engineering tasks requires a structured self-assessment approach. This involves categorizing task complexity and applying specific improvement plans to refine previous outputs.

You should avoid using a self-reflection framework for code review when the task is trivial and requires no architectural alignment or dependency checks. In such cases, a deep complexity triage is unnecessary and adds overhead to simple outputs.