ada-code-efficiency-review

Run an efficiency-focused review that reads every changed file in full, then checks an 8-category checklist covering redundant computation, N+1 patterns, missed concurrency, hot-path bloat, TOCTOU races, memory issues, excessive reads, and silent failures. Each finding includes a concrete fix and severity grading.

A general code review checks security, logic errors, and test regressions before commit. An efficiency review targets waste only: code that works correctly but does too much work, allocates too often, or leaks resources over time.

Look for loop bodies calling FirstOrDefault or Find on the same collection each iteration, per-event linear scans, or inner queries that a pre-built dictionary lookup could satisfy. Flag them only when N can grow large or the call sits on a hot path.

Yes, the 8-category checklist is language-agnostic and applies to any codebase. The skill includes extra Blazor WASM-specific guidance for JS interop batching, render-cycle allocations, and await re-entrancy, but the core audit works anywhere.

Skip it for style-only changes, documentation, and config tweaks, or when the user only wants bug detection rather than waste analysis. It is most valuable on diffs with 3+ changed files or changes touching hot paths like startup and render loops.

A changed line can look fine in isolation but be called from a hot loop elsewhere in the file. Reading full files and tracing cross-file dependencies like interop layers and service registrations prevents false negatives where leaks hide outside the diff.