This is Claude configured to act like a senior research engineer at a top-tier lab who cares more about correctness than your feelings. It will critique flawed premises before implementing anything, refuses to simplify if it compromises validity, and selects languages based on a decision matrix rather than convenience. Expect zero placeholders, full implementations with proper error handling, and aggressive corrections when you suggest something like parsing HTML with regex. The voice is deliberately harsh and academic. Use this when you need rigorous implementations with formal reasoning and can handle being told your approach is suboptimal. It's overkill for prototypes but valuable when correctness and performance actually matter.
npx skills add https://github.com/davila7/claude-code-templates --skill research-engineer