Individual skill

perf

Make a studio's game faster with a measured loop in real GPU Chrome on a computer and an emulated phone. Measure frame times, startup, downloads, memory and netplay for both server-room replicas or a browser-host pair. Profile hot functions, try one small change, compare alternating runs, and keep it only when the improvement exceeds noise, guarded measures hold and the two-browser check passes. Write the report in perf/. Use for stutter, lag, slow loading, phone drain, rendering costs or an overnight performance pass.

plugins/homie-rocks/homie/skills/perf/SKILL.md

Read pinned source

Source commit d780783eba4ecc7c8b9429f95c9362df95f07c61

What was checked

Matches the observed repository HEAD
Yes
Scanner rules are current
Yes
Repository scan complete
No
Skill scan complete
No
Observed behavior
Network or broad shell
Last checked
Nov 4, 2026

These are observations at the recorded commit, not a guarantee of safety. Review the source before running it.

Install this skill

Choose an agent in your Stack or copy a command below. Commands fetch the pinned commit and copy only this skill directory. Existing destinations are refused. Copying a command does not install anything.

An install command is unavailable: current rules, complete repository and skill scans, an eligible behavior level and a permitted license are required.

Repository context

Repository stars (not skill ratings)
1,537
Repository last updated
Nov 7, 2026
License
Apache-2.0

Observed agent integration: No specific integration evidence. Standard SKILL.md installation does not imply verified compatibility.

Files and excerpt

File metadata and permitted, redacted excerpts come from the recorded scan. Source links are pinned to that commit.

Browse files
Read excerpt

# Make a game faster, honestly

The question is never "does it feel faster". It is: **is this build measurably better than the one before it, beyond
the noise, without looking or playing any differently, and does it still work?** Every number here comes from a real
Chrome on this computer's GPU, two browsers in a room playing the same seeded presses every run. New games use server rules by default. When the server hosts the rules, both browsers are replicas (`replica` and `replica-2`); reports and regression guards include both, and the default goal uses `replica` instead of `host`. An explicit `--goal` keeps the role you name.

One script: `scripts/perf.mjs` in this skill's folder (Claude Code:
`node "${CLAUDE_PLUGIN_ROOT}/skills/perf/scripts/perf.mjs" <command>`), run from inside the studio. It drives the
studio's own `@homie-rocks/studio` (0.19.0 or later: `homie-studio perf`, `perf compare`, `check`, `build --maps`;
0.19.1 also reads whether each big script is minified),
opens at most two browsers at a time, all muted, and writes everything it measures under `.perf/<game>/`
(git-ignored). It prints paths and a few numbers, never the data: read the files it names.

## 0. Before anything