A new Claude Code skill launched on GitHub addresses a critical challenge in AI-assisted development: maintaining genuine skill acquisition while using code generation tools. The Learning Opportunities skill, created by Dr. Cat Hicks and Dr. Michael Mullarkey, offers optional learning exercises based on cognitive science research to help developers build expertise rather than just velocity.
The skill gained 141 points and 27 comments on Hacker News within days of its May 14, 2026 release, reflecting strong developer interest in deliberate learning practices.
Five Specific Cognitive Risks Addressed
The tool targets documented problems in AI-assisted coding:
- Generation effect: Accepting generated code without creating your own diminishes active processing and memory consolidation—research shows that information learners create sticks better than information they read
- Fluency illusion: Clean AI code can feel more understood than it truly is—reading comprehension doesn't equal implementation ability
- Spacing effect: High machine velocity encourages cramming rather than spaced learning, which research shows is critical for long-term retention
- Metacognition gaps: Fast workflows leave little room to monitor your own learning progress and knowledge gaps
- Retrieval practice loss: Complete AI answers reduce beneficial self-testing, which strengthens memory more than re-reading
How the Skill Works
After completing significant architectural work—creating new files, refactoring, or implementing unfamiliar patterns—Claude asks if you'd like a quick learning exercise on the relevant topic. If accepted, Claude deliberately pauses and waits for your input rather than providing answers, encouraging active mental effort. Exercises take 10-15 minutes.
Six Exercise Types Based on Learning Science
The skill offers varied approaches to deliberate practice:
- Prediction → Observation → Reflection: Predict what code will do, run it, reflect on differences
- Generation → Comparison: Write your own implementation, compare with AI version
- Trace the path: Step through execution mentally
- Debug this scenario: Find and fix intentional bugs
- Teach it back: Explain concepts in your own words
- Retrieval check-ins: Recall key concepts without looking
Optional and Non-Intrusive Design
The skill respects that developers sometimes need pure velocity for deadlines, exploratory coding, or simple tasks. It offers structured learning opportunities only when beneficial, making it entirely optional.
Research Foundation
Dr. Hicks, a psychological scientist studying software teams, and Dr. Mullarkey, a machine learning engineer with research background, developed the tool based on established learning science principles including desirable difficulties, the testing effect, spaced repetition, metacognitive monitoring, and the generation effect.
Key Takeaways
- Learning Opportunities skill launched May 14, 2026, receiving 141 points on Hacker News
- Addresses five cognitive risks: generation effect, fluency illusion, spacing effect, metacognition gaps, and retrieval practice loss
- Offers six types of optional 10-15 minute learning exercises after significant coding work
- Created by Dr. Cat Hicks and Dr. Michael Mullarkey based on cognitive science research
- Non-intrusive design respects when developers need pure velocity versus deliberate learning