Claude Code Adds Agent View, Gamifies Collaborative Coding

Researchers and developers have published a Claude Code and Codex plugin—learning-opportunities—that offers brief, evidence-based learning exercises tied to coding work to help users build durable expertise rather than only producing code. The skill proposes 10–15 minute tasks (prediction, generation, retrieval practice, spaced repetition) after architectural changes or commits, and integrates with an orient plugin to generate repo-specific orientation lessons. It supports Codex and Claude Code plugin marketplaces and an optional auto mode to trigger exercises post-commit. The project aims to counter AI-driven learning harms like the generation effect, fluency illusion, and reduced metacognition by embedding science-backed study techniques into developer workflows. Installation and usage instructions are provided.

A new Claude Code and Codex plugin, learning-opportunities, offers an evidence-based “dynamic textbook” skill that proposes short (10–15 minute) learning exercises tied to coding actions like new files, schema changes, or refactors. Developed as a Claude Code and Codex marketplace plugin, it uses techniques from cognitive science — prediction, generation, retrieval practice, spaced repetition, and MCII goal-setting — to help developers build durable expertise rather than just accepting generated code. Optional components include an auto-trigger post-commit hook, an orient tool that generates repo orientation lessons based on program-comprehension research, and integration instructions for installation and local development. The project aims to counteract AI-driven fluency illusions and promote deliberate learning during AI-assisted coding.