tranhieutt/perf-profile

name: perf-profile description: "Identifies performance bottlenecks, measures against budgets, and generates prioritized optimization recommendations. Use when investigating slow performance or when the user mentions performance profiling, bottleneck, or optimization." argument-hint: "[system-name or 'full']" user-invocable: true allowed-tools: Read, Glob, Grep, Bash context: fork agent: performance-analyst effort: 3 when_to_use: "Use when investigating slow performance, identifying bottlenecks, or when the user mentions performance issues, profiling, latency, or optimization targets."

When this skill is invoked:

1. Determine scope from the argument:

   • If a system name: focus profiling on that specific system
   • If full: run a comprehensive profile across all systems

2. Read performance budgets — Check for existing performance targets in design docs or CLAUDE.md:

   • Target FPS (e.g., 60fps = 16.67ms frame budget)
   • Memory budget (total and per-system)
   • Load time targets
   • Draw call budgets
   • Network bandwidth limits (if multiplayer)

3. Analyze the codebase for common performance issues:

   CPU Profiling Targets:

   • _process() / Update() / Tick() functions — list all and estimate cost
   • Nested loops over large collections
   • String operations in hot paths
   • Allocation patterns in per-frame code
   • Unoptimized search/sort over data entities
   • Expensive physics queries (raycasts, overlaps) every frame

   Memory Profiling Targets:

   • Large data structures and their growth patterns
   • Texture/asset memory footprint estimates
   • Object pool vs instantiate/destroy patterns
   • Leaked references (objects that should be freed but aren't)
   • Cache sizes and eviction policies

   Rendering Targets (if applicable):

   • Draw call estimates
   • Overdraw from overlapping transparent objects
   • Shader complexity
   • Unoptimized particle systems
   • Missing LODs or occlusion culling

   I/O Targets:

   • Save/load performance
   • Asset loading patterns (sync vs async)
   • Network message frequency and size

4. Generate the profiling report:

   ## Performance Profile: [System or Full]
   Generated: [Date]
   
   ### Performance Budgets
   | Metric | Budget | Estimated Current | Status |
   |--------|--------|-------------------|--------|
   | Frame time | [16.67ms] | [estimate] | [OK/WARNING/OVER] |
   | Memory | [target] | [estimate] | [OK/WARNING/OVER] |
   | Load time | [target] | [estimate] | [OK/WARNING/OVER] |
   | Draw calls | [target] | [estimate] | [OK/WARNING/OVER] |
   
   ### Hotspots Identified
   | # | Location | Issue | Estimated Impact | Fix Effort |
   |---|----------|-------|------------------|------------|
   | 1 | [file:line] | [description] | [High/Med/Low] | [S/M/L] |
   | 2 | [file:line] | [description] | [High/Med/Low] | [S/M/L] |
   
   ### Optimization Recommendations (Priority Order)
   1. **[Title]** — [Description of the optimization]
      - Location: [file:line]
      - Expected gain: [estimate]
      - Risk: [Low/Med/High]
      - Approach: [How to implement]
   
   ### Quick Wins (< 1 hour each)
   - [Simple optimization 1]
   - [Simple optimization 2]
   
   ### Requires Investigation
   - [Area that needs actual runtime profiling to determine impact]
   

5. Output the report with a summary: top 3 hotspots, estimated headroom vs budget, and recommended next action.

Rules

• Never optimize without measuring first — gut feelings about performance are unreliable
• Recommendations must include estimated impact — "make it faster" is not actionable
• Profile on target hardware, not just development machines
• Distinguish between CPU-bound, GPU-bound, and I/O-bound bottlenecks
• Consider worst-case scenarios (maximum entities, lowest spec hardware, worst network conditions)
• Static analysis (this skill) identifies candidates; runtime profiling confirms

Protocol

• Question: Reads system name or full from argument
• Options: Skip
• Decision: Skip
• Draft: Profile report shown in conversation only
• Approval: Skip — read-only; no files written by default

Output

Deliver exactly:

• Top 3 hotspots — system, estimated impact, and recommended fix
• Budget headroom — actual vs target for each metric in technical-preferences.md
• Optimization priority list — ranked by impact/effort ratio
• Next action: PROFILE CONFIRMED (static analysis sufficient) or RUNTIME PROFILING NEEDED (list what to measure)