Reaction
Knowledge gaps in large codebases, with or without AI
A developer says they were unfamiliar with one detail of how tagged skills attach to prompts passed to the Claude SDK. They argue that working with partial understanding is normal in engineering.
TLDR
A developer pushes back on criticism of that knowledge gap, saying the codebase has over 300 contributors and millions of lines of code. They argue that no one knows every implementation detail in a codebase that size, regardless of AI use, and worry that treating such gaps as failures could hold junior engineers back.
Combined views
—
1 Source, first seen ago
