agentleFS
Sign inStart free
ThiagoFdaSLopesThiagoFdaSLopes/game-dev-skills/skills/game-optimize/SKILL.md

game-optimize

Skill2 starsChanged 24 days agoMITView on GitHub

Diagnose and fix FPS drops, stutter, memory growth, long loads, input latency or scaling problems by measuring first, changing one bottleneck at a time, and comparing before/after under the same conditions. Use when the user mentions performance, frame rate, hitching, memory, load time, GC spikes, draw calls or latency. Without measurements it proposes an instrumented experiment instead of promising gains.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. AirMile/claude-config★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository6 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read before starting: references/principles.md, references/project-profile.md and the project profile itself, references/question-format.md.

    line 11
  2. Distinctive:

    Baseline: - run the profiler, a frame-time capture, memory snapshot or a scripted benchmark if the environment allows.

    line 16
  3. Distinctive:

    Iterate or stop: - if the target is met, stop.

    line 20
  4. Distinctive:

    Ask one shared-format question only when the fix trades off something the user must weigh (visual quality vs frame time, determinism vs …

    line 24

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

game-optimize

Measure, hypothesise, change one thing, measure again. Never promise a number you have not seen.

Read before starting: references/principles.md, references/project-profile.md and the project profile itself, references/question-format.md. Budgets already recorded by the project are the targets, and the measuring tools it lists (profile: Checks) come before any new instrumentation.

Process

  1. Target - platform, target frame time/memory, and the symptom in one line (stutter every ~2s on Android mid-range, target 16.6ms). Pull platform and quality settings from the project.
  2. Baseline - run the profiler, a frame-time capture, memory snapshot or a scripted benchmark if the environment allows. If it cannot run here, hand the user the exact instrumentation to add or the profiler steps to follow, then wait for numbers. Never estimate a baseline.
  3. Hypothesis - name the suspected bottleneck and why the evidence points there. Usual suspects: per-frame allocations and GC, physics tick vs render rate, overdraw and draw-call count, unbatched UI, synchronous asset loads, string/boxing in hot loops, pathfinding without budgets, netcode serialization per tick.
  4. Change one thing - the smallest change that tests the hypothesis. Preserve gameplay feel and correctness; a faster game that plays differently is a regression.
  5. Compare - same scene, same inputs, same build config, same device. Report both numbers with units.
  6. Iterate or stop - if the target is met, stop; do not gold-plate. If not, next hypothesis.

Fork

Ask one shared-format question only when the fix trades off something the user must weigh (visual quality vs frame time, determinism vs throughput, memory vs load time). Otherwise proceed.

Closing

Target:        platform + budget
Baseline:      measured value(s) + how captured, or "not measured: <reason>"
Change:        what was changed and why
After:         measured value(s) under identical conditions
Verdict:       met / not met / unmeasured
Open risks:    feel or correctness side effects to check; other suspects

Never

  • report a gain that was not measured under comparable conditions
  • optimise a path the profile does not implicate
  • change gameplay behaviour as a side effect without naming it

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free