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/plausibleventures/lattice/determinismnpx skills add plausibleventures/lattice --skill determinismgit clone --depth 1 https://github.com/plausibleventures/latticeWrote 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/plausibleventures/lattice/determinism)<a href="https://agentmods.dev/skills/plausibleventures/lattice/determinism"><img src="https://agentmods.dev/badge/skills/plausibleventures/lattice/determinism.svg" alt="Measured on agentmods" 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 | $0.00077 | $0.02919 |
| Opus 5 | $0.00039 | $0.01460 |
| Sonnet 5 | $0.00015 | $0.00584 |
| Haiku 4.5 | $0.00008 | $0.00292 |
Grade A, and why
determinism 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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 233 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Determinism
The claim this kit makes is that a session can be replayed from a seed and an input log and land on the same pixel. That is a real property with real edges, and most of the ways it breaks are things nobody would guess.
Two tiers, because the language only promises one
ECMA-262 specifies + - * /, Math.sqrt, Math.imul and the bitwise operators exactly. It
explicitly does not require sin, cos, pow, exp or log to be correctly rounded, so two
conforming engines may disagree in the last bit.
| arithmetic | promise | may reach | |
|---|---|---|---|
| Tier A | + - * /, sqrt, imul, bitwise, abs/min/max/floor/round |
bit-identical on every engine | hashes, save files, replays, anything |
| Tier B | sin, cos, atan2, pow, exp, log |
correct to within an ulp or so | pixels only — never hashed, never persisted |
Tier B is not banned; a cost curve is b · r^k and there is no honest way around that. What it
must not do is reach a save file or a hash.
Consequences you will meet:
- There are no sine or expo easings anywhere in the kit. A tween drives a position, a position gets written to a save, and the save no longer replays.
- Pick a dyadic offline exponent.
0.5,0.625and0.75are computed as a chain ofMath.sqrtand multiplies, so credited time is Tier A for free.0.6is three per cent stingier than0.625and a whole determinism tier worse. - Path costs are integers — the octile metric
14·min(dx,dy) + 10·|dx−dy|, exact and admissible with nosqrt. Float summation is associative only by luck, so two engines can pop equal-cost nodes in a different order and return different — both optimal, both different — routes. SEMITONEexists so a game need not callpowfor a musical interval.
The one place the tier rule bites hardest is a field that feeds itself. A height field being
eroded is state that feeds the next step, so a last-bit disagreement in one pow is amplified by
the next droplet that steers on the gradient it perturbed — and after a hundred thousand
iterations two conforming engines have the river in a different valley. In a loop like that, use
only Tier A arithmetic and do not bother tagging anything as presentation, because every value in
the file reaches the next iteration.
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.
- 4d ago First seen · 233 lines · 77 tokens per session scan A e265ffd3c8e6
determinism is a skill published in the GitHub repository plausibleventures/lattice (35 stars, last pushed 12d ago), licensed MIT. It adds 77 tokens to every session and 2,919 once invoked, about $0.0004 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 2026-08-30.
Other skills, from other repositories
phaser3-engineer
!cat skills/shared/game-visual-foundations.md 2>/dev/null || echo "=== Visual Foundations not loaded ===" !cat skills/shared/protocols/ux-protocol.md 2>/dev/null || true !cat skills/shared/protocols/input-validation.md 2>/dev/null || true !cat skills/shared/protocols/tool-efficiency.md 2>/dev/null || true !cat…
dedicated-server
Use when building dedicated servers — headless export, server architecture, lobby management, and deployment.
dialogue-manager
Use when using the Dialogue Manager addon — .dialogue files with titles, responses, conditions and mutations, runtime balloons, and C# support.
event-bus
Use when implementing decoupled communication between nodes — global EventBus autoload with typed signals.
godot-optimization
Use when optimizing Godot games — profiler, draw calls, physics tuning, memory management, and common bottlenecks.
limboai
Use when using the LimboAI addon — behavior trees and hierarchical state machines (C++ GDExtension) with a visual editor, BTTask subclassing, and a blackboard.