A new GitHub repository called learning-opportunities introduces a novel approach to maintaining developer skills while using AI coding assistants. Created by DrCatHicks, this tool integrates with both Claude Code and Codex to offer evidence-based learning exercises during development sessions. Rather than passively accepting generated code, developers can engage in active learning opportunities that interrupt the typical AI-assisted workflow.

The system triggers optional exercises after completing significant architectural work like creating new files, schema changes, or refactors. These exercises employ techniques from learning science including prediction, generation, retrieval practice, and spaced repetition. Developers can install the plugin through their respective AI assistant's marketplace, with additional auto-prompting capabilities available for git commits.

Beyond addressing the fluency illusion where clean generated code masks incomplete understanding, the tool counters risks like reduced generation effect and spacing violations. By forcing developers to pause and actively engage with their work—predicting outcomes, explaining concepts, or debugging—the system promotes genuine comprehension over mere code completion.

Installation requires adding the marketplace URL to Claude Code or Codex, then installing the learning-opportunities plugin. An optional orient skill helps newcomers navigate unfamiliar repositories using research-backed comprehension strategies.