Skip to content
Back to skills

Rng Seed

ASecurity

Use when procedural generation, combat rolls, loot, replays, rollback, or deterministic tests require reproducible random streams that remain stable across refactors and resimulation.

  • 11 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added October 23, 2026
ai-agentsgoaws

Security analysis

A100/100

Scanned October 23, 2026

npx -y skills add jammyfu/open-game-skills --skill rng-seed --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Rng Seed?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Rng Seed
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/jammyfu-rng-seed/badge)](https://www.skillsdirectory.com/skills/jammyfu-rng-seed)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: rng-seed
description: Use when procedural generation, combat rolls, loot, replays, rollback, or deterministic tests require reproducible random streams that remain stable across refactors and resimulation.
---

# RNG seed

A seed is only one part of a reproducibility contract.

## Stream contract

Publish the RNG algorithm/version plus stable named streams for domains whose call order must not affect each other, for example `world`, `combat`, `loot`, `cosmetic`. Derive child stream seeds from stable semantic identifiers, not mutable container iteration order.

Each stream has serializable state or a counter/index that can be snapshotted. A replay or rollback snapshot restores the exact stream state/counter for the rewind tick; reseeding from a root seed and hoping the same number of calls occurred is not sufficient.

Random draws happen on the logical simulation path that owns the outcome. Render-only variation uses a separate presentation stream and must not mutate gameplay RNG.

## Replay identity

A reproducible replay records at least the game/simulation version, RNG algorithm/version, root/run identity, initial deterministic state/config, named stream derivation rules, logical tick/input trace, and any authoritative external events. If rules or algorithm change, treat old replays as a different compatibility version rather than silently producing new outcomes.

## Acceptance

Run the same recorded initial state + inputs twice and compare deterministic state hashes and every named RNG stream state at checkpoints. Insert an unrelated cosmetic random draw and verify combat/loot results do not change. Snapshot mid-run, restore, and reproduce subsequent draws exactly. Unrun engine/platform replays stay `not-run`.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

View all in ai-agents →

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.