Question 6
Domain 1A team writes a PreToolUse hook for *every* coding-style preference (indent width, brace style, naming conventions). What's the structural critique?
Correct answer: A
Explanation
PreToolUse hooks are meant to enforce trust boundaries, not every preference. The answer fits because style rules are usually advisory, so putting "indent width, brace style, naming conventions" into hooks is "over-engineered"; those belong in CLAUDE.md, which is the right place for guidance without adding friction and complexity.
Why each option is right or wrong
A. Over-engineered for the trust requirement. Style preferences are advisory; CLAUDE.md is the right layer. Hooks add friction and complexity for nudges that don't need enforcement.
Under the hook model, PreToolUse is the enforcement layer for actions that need gating before a tool runs, not a general-purpose style system. The cited critique is that coding-style preferences are advisory rather than trust-boundary controls, so routing every preference through a hook adds unnecessary operational overhead and complexity instead of simply documenting them in CLAUDE.md, which is the intended guidance layer.
B. PreToolUse hooks can't read tool input.
PreToolUse exists to inspect or gate tool use, so this absolute claim does not fit.
C. Style hooks slow down each tool call to the point of timing out.
Performance cost may exist, but style hooks are not inherently guaranteed to cause timeouts.
D. Style hooks bypass the model and produce non-LLM-aware fixes.
Hooks are part of the orchestration layer, not a mechanism that inherently bypasses model awareness.