joaoariedi

Use when analyzing performance bottlenecks or optimizing critical code paths. Finds N+1 queries, blocking I/O, memory leaks, and algorithmic complexity. Read-only analysis with optional benchmarking via Bash.

Use when investigating bugs, test failures, or unexpected behavior. 4-phase root cause investigation before attempting any fix. Enforces the Iron Law: no fixes without root-cause investigation first. Prevents shotgun debugging and cargo-cult fixes.

Use when estimating the effort, complexity, or risk of a proposed code change or an existing diff. Computes a deterministic Contribution Complexity score from version-control metadata, then flags AI-native risk. Read-only analysis.