performance-profiling

Profile likely hotspots and turn performance work into measurable next steps.

Performance Profiling

Trigger Signals

  • the task is slow, resource-heavy, or performance-sensitive

Inspect First

  • hotspot code paths, timing data, memory cues, and benchmark surfaces

Workflow

  1. identify the slow path and measurement surface
  2. isolate likely bottlenecks
  3. define low-risk experiments and verification
  4. report findings and next action

Output Contract

  • hotspot summary
  • likely bottlenecks
  • measurement or benchmark plan
  • recommended next step

Failure Modes

  • no measurable baseline exists

Escalation

  • escalate when optimization risks correctness or public contracts