This orchestrates OpenAI's o4-mini-deep-research model to handle comprehensive research tasks with web search. It's clever about prompt enhancement: if you throw it something brief like "latest AI trends," it asks 2-3 clarifying questions about scope and depth before running the actual research. Saves everything to timestamped files for reproducibility. The workflow is token-efficient because deep research takes 10-20 minutes, so it just blocks and waits rather than polling for status updates. Best for when you need actual investigation with sources, not just a quick answer. The technical versus general research detection is a nice touch for tailoring follow-up questions.
npx skills add https://github.com/glebis/claude-skills --skill deep-research