# Game Performance Budget

> Use when defining or reviewing game frame-time, memory, loading, network, CPU, GPU, allocation, or thermal budgets against captured measurements and target hardware tiers.

- Skill: `hoatv2211/game-performance-budget` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add hoatv2211/game-performance-budget`
- Raw SKILL.md: https://api.skillmd.com/api/skills/hoatv2211/game-performance-budget/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Finance & Business
- License: MIT
- Author: hoatv2211 (https://skillmd.com/u/hoatv2211)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/hoatv2211/game-performance-budget

---

# Game Performance Budget

## Overview
Compare reproducible performance measurements with explicit budgets instead of treating a fast developer machine or average frame rate as proof.

## When to use
Use for milestone gates, optimization reviews, target-hardware qualification, regression checks, memory reviews, loading tests, and network performance budgets.

## When NOT to use
Do not use to guess performance without captures, to optimize before a measured bottleneck, or to certify hardware that was not tested.

## Required inputs and context discovery
Collect build ID, scene or scenario, target hardware tier, quality settings, warm-up method, capture duration, measurement tool, percentile metrics, memory categories, loading phases, network conditions, and thresholds.

## Safety and risk level
Read-only analysis. Profiling unavailable hardware or builds is `BLOCKED`; estimates may be `Unverified` but never PASS.

## Workflow
1. Define budgets per target hardware tier and gameplay scenario.
   Completion criterion: units, thresholds, and allowed variance are explicit.
2. Validate capture identity and measurement method.
   Completion criterion: build, settings, warm-up, duration, and profiler source are recorded.
3. Compare observed percentiles and peaks with each budget.
   Completion criterion: every metric reports target, observed value, delta, and verdict.
4. Attribute exceeded budgets to measured subsystems or mark attribution unresolved.
   Completion criterion: bottleneck claims cite profiler evidence rather than intuition.
5. Define optimization experiments and regression gates.
   Completion criterion: each action includes a metric, target, owner, and retest scenario.

## Evidence and output contract
Produce `performance-budget-report.json` with build and hardware identity, scenarios, targets, observations, deltas, verdicts, profiler artifacts, limitations, and retest actions.

## Handoff contract
Record failing tiers, worst scenarios, capture paths, suspected subsystems, known measurement noise, owners, and exact retest commands or steps.

## Pitfalls and anti-rationalization
- Average FPS hides spikes; use percentiles and peaks.
- Editor measurements do not certify player builds.
- One hardware tier cannot stand in for another.
- Missing captures remain `BLOCKED`, not estimated PASS.

## Verification checklist
- [ ] Build, hardware, settings, and scenarios are exact.
- [ ] Budgets include units and thresholds.
- [ ] Observations cite profiler artifacts.
- [ ] Percentiles and peaks are reported where relevant.
- [ ] Retest actions preserve the same measurement method.

## References and scripts
Use the bundled [scripts/performance_budget.py](scripts/performance_budget.py) for normalized budget comparisons and project profilers for captures. The helper does not launch or profile a live build.