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 agentmods add skills/jangles-byte/atelier/performance-craftnpx skills add jangles-byte/atelier --skill performance-craftgit clone --depth 1 https://github.com/jangles-byte/atelierWhat 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 | $0.00146 | $0.00552 |
| Opus 5 | $0.00073 | $0.00276 |
| Sonnet 5 | $0.00029 | $0.00110 |
| Haiku 4.5 | $0.00015 | $0.00055 |
Grade A, and why
performance-craft 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 2d ago.
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.
What it actually says
Performance Craft
A dropped frame is a design defect the user feels before they can name it. This skill treats the frame budget as a design material: know the budget, spend it on what the philosophy values, and degrade in the order that does the least visual damage.
The budget
Frame time = 1000 / refresh rate: 8.3ms at 120Hz, 16.6ms at 60Hz — minus the platform's own compositing overhead. Design for the display the audience has: ProMotion phones and most gaming displays are 120Hz; never hardcode 60. Measure before and after any motion work — profilers, not vibes (DevTools Performance / Instruments' Animation Hitches / engine profilers).
Which reference to load
| Situation | Load |
|---|---|
| Web: CSS/JS/canvas/WebGL animation, scroll jank, layout thrash | references/web-rendering.md |
| Apple: SwiftUI/UIKit hitches, Metal, ProMotion, battery/thermals | references/apple-silicon.md |
| Game engines, particles at scale, and the degradation ladder | references/budgets-and-degradation.md |
Non-negotiables
- Animate compositor-friendly properties (transform, opacity — and their platform equivalents) unless a measured reason says otherwise.
- Timing from delta-time or the platform's frame callback, never a hardcoded 1/60 assumption.
- Always-on ambient animation must be provably cheap: paused off-screen, GPU-only, and off the performance cores / not preventing CPU idle.
prefers-reduced-motion(and platform equivalents) is also the low-power path — wire them together.- When the budget is blown, cut by the degradation ladder — never ship the stutter, and never fix it by cutting the design's primary feedback motion first.
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 2d ago First seen · 41 lines · 0 tokens per session scan A 7a5adf336f30
performance-craft is a skill published in the GitHub repository jangles-byte/atelier (2 stars, last pushed 1mo ago), licensed MIT. It adds 146 tokens to every session and 552 once invoked, about $0.0007 per session on Opus 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 2026-08-31.
Other skills, from other repositories
gamified-app
A multi-frame gamified mobile-app prototype — three phone frames on a dark showcase stage. Frame 1: cover / poster, Frame 2: today's quests with XP ribbons and a level bar, Frame 3: quest detail. Vivid quest tiles, level ribbon, bottom tab bar. Use when the brief asks for a "gamified app", "habit tracker", "RPG-style…
html-ppt-zhangzara-8-bit-orbit
A gamer's journey building a retro-arcade collection — the obsession, the hunt, and what the machines came to mean. Built as a decision-grade story deck for friends, hobby community.
state-trees
Create, inspect, and edit StateTree assets — states, tasks, transitions, conditions, considerations, and property bindings for AI behavior and game logic (StateTreeService). Use when the user asks to build a StateTree, add states/tasks/transitions, bind task properties, set up Utility AI considerations, or wire…
animation-blueprint
Navigate, inspect, and author Animation Blueprints — state machines, states, transitions, and transition rules (AnimGraphService). Use when the user asks to create or edit an Animation Blueprint/AnimBP, add a state machine or states, wire transitions between states, set a state's animation, or inspect AnimGraph…
animation-montage
Create and manipulate Animation Montages — sections, slots, segments, branching points, notifies, and blend settings (AnimMontageService). Use when the user asks to create a montage (from an animation or empty), add or link montage sections, set up slots/segments, add branching points, or adjust montage blend timing.
animsequence
Create and edit AnimSequence keyframes and tracks — add/remove keyframes, set bone tracks, query sequence info, and bake edits (AnimSequenceService). Use when the user asks to create a new animation sequence, add keyframes/tracks, or edit raw animation data. For constraint-aware bone posing, also load…