reflect

Mid-task reflection uses a three-step workflow—task-adherence, information-completeness, and completion-check—to evaluate progress, identify blockers, and reaffirm alignment before making major changes. It surfaces actionable next steps and optionally stores insights in memory.

Metacognition in software engineering involves pausing to evaluate your own task progress and information gathering. Use a reflection checkpoint during ongoing research or development when uncertainty arises or before large changes to reduce drift and improve decision quality.

Task adherence and information completeness are checked through a guided reflection sequence that enforces evaluating whether your current actions match the goal and whether you have gathered enough context before moving forward with large changes.

Yes, the reflection workflow includes optional memory read and write operations using Serena tools. This allows you to surface past context and decisions from memory, and store new insights to maintain better alignment throughout your task.

Preventing scope drift is best achieved through structured self-reflection prompts that enforce task-adherence and completion-checks. By pausing to evaluate information completeness and progress blockers, you reduce drift and improve decision quality before proceeding.

No specific dependencies or environment setup are required. The reflection workflow operates independently using Serena metacognitive tools to guide the three-step task-adherence, information-completeness, and completion-check sequence without external components.