Estella is a 2D and 3D game engine with a TypeScript interface, a C++/WebAssembly core, and a visual editor that also acts as an MCP server. It helps developers create, test, and export games for the web, desktop, mini-game platforms, playable ads, and native mobile devices. Its catalogue add-ons let coding agents operate the editor through the same tool surface used by its interface.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add esengine/estella --skill estella-performancegit clone --depth 1 https://github.com/esengine/estellaWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/esengine/estella/estella-performance)<a href="https://agentmods.dev/skills/esengine/estella/estella-performance"><img src="https://agentmods.dev/badge/skills/esengine/estella/estella-performance/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/esengine/estella/estella-performance"><img src="https://agentmods.dev/badge/skills/esengine/estella/estella-performance.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00088 | $0.01219 |
| Opus 5.5 | $0.00035 | $0.00488 |
| Sonnet 5.5 | $0.00018 | $0.00244 |
| Haiku 4.5 | $0.00009 | $0.00122 |
Grade A, and why
estella-performance scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured today.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Performance
Never say a game is fast or slow without a number, and never fix what was not measured. The loop is: measure, find the one largest cost, change one thing, measure again the same way.
Measure honestly
- Measure the RUNNING game:
set_play { state: "playing" }, thenstepa few hundred frames so shaders, streaming and generation are warm. The first frames of anything are not its cost. profile_framessays where the frame went:cpuMs(work),gpuMs, andwaitMs, which is the CPU waiting for the display and NOT a hotspot.stalled: truemeans no frame ran: say so, do not read the zeroes as fast. An editor window in the background runs at about 1 FPS, and that is not a result either.- Read p95 and p99 and the long frames, not only the average: a stutter is one bad frame the average hides.
- This machine is not the target. A phone's CPU is several times slower than a desktop's: a system costing 0.5 ms
here is 2 or 3 ms of a phone's 16. Judge desktop numbers at that scale, and measure on the device when it matters:
a development build (
export_game { debugChannel: true }) appears indevice_targets, thenset_profiler_source { source: realm }pointsprofile_framesandcapture_frameat it. A game can also record a.esprofwithProfileRecorderon a player's device;profile_capturereads it,worstFramefirst. - Measure before and after the same way: same scene, same camera, same frames stepped. Two runs differ by noise; repeat and take the better of each before believing a small gain.
CPU: which system
profile_frames lists cost per domain (scripts, render, physics, ui) and the costliest systems with what their
queries walked. get_systems gives the same per frame.
filterednearscannedis a system walking the whole world to throw it away. Query for the component it needs, or use a change filter that skips, not one that visits.- A system that runs every frame for a feature the scene may not use is gated with
activeWhen:app.addSystemToSchedule(schedule, system, { activeWhen: [MyComponent] })skips it while no entity has the component. - A frame where nothing changed should cost almost nothing. Writing a component every frame with the value it already has makes everything downstream redo its work (a UI write lays the whole tree out again): write only on change.
- Allocation in a per-frame loop (new objects, arrays, closures) comes back as garbage-collection long frames.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- today First seen · 69 lines · 88 tokens per session scan A 794c93843634
estella-performance is a skill published in the GitHub repository esengine/estella (1,174 stars, last pushed today), licensed Apache-2.0. It adds 88 tokens to every session and 1,219 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-11-19.
estella-performance compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| estella-performanceesengine/estella | Skill | 1.2k | today | Apache-2.0 | A | 88 | Claude Code, Codex |
| cougr-game-startersalazarsebas/Cougr | Skill | 53 | 28d ago | MIT | A | 68 | Claude Code, Codex |
| hearth-artechoo19/hearth | Skill | 43 | 1mo ago | MIT | A | 89 | Claude Code |
| hearth-feelechoo19/hearth | Skill | 43 | 1mo ago | MIT | A | 97 | Claude Code |
| hearth-buildechoo19/hearth | Skill | 43 | 1mo ago | MIT | A | 75 | Claude Code |
| hearth-codeechoo19/hearth | Skill | 43 | 1mo ago | MIT | A | 88 | Claude Code |
| hearth-designechoo19/hearth | Skill | 43 | 1mo ago | MIT | A | 109 | Claude Code |