Learning

1. Identify the topic and the specific insight
2. Write the insight as a concise, actionable statement (e.g., "Always check X before doing Y")
3. Record the context (what task or situation prompted this)
4. Append to the learning store via the learning_save tool or by updating the reflection file

Reviewing Learnings

1. Load the current learning store
2. Filter by topic if the user has a specific area in mind
3. Present the most relevant or recent entries
4. Suggest which learnings might apply to the current task

Surfacing Lessons Before a Task

Before starting a complex task, check the learning store for entries tagged with a related topic. Present any relevant lessons as a "heads up" before proceeding.

Learning Types

Key distinction: gap = "I don't know X", feature_request = "Flopsy can't do X yet". Use error instead of gap when there's an actual failure output to record.

Guidelines

• Learnings should be specific and actionable, not vague observations
• Limit entries to a manageable number; periodically review and consolidate
• Tag entries with topics so they can be retrieved contextually
• The learning system is a complement to skills, not a replacement; new learnings should be promoted to skill documentation when they become general best practices