Pick a die, choose how many, and roll. The stage looks like a dice surface before the first roll without pretending a random result already exists.
Inspect the current pool, probability model, duplicate/replacement rules and result constraints with a separate browser-local verification layer. It does not choose or alter the primary result.
Use the broad picker, number, wheel, name, team or dice workflow first. Move to a specialist only when the probability model or output structure genuinely changes.
Run repeatable tabletop/classroom expressions such as 2d20kh1+5. Individual rolls stay visible so keep-highest/lowest decisions are auditable.
| Expression | Raw dice | Kept | Total |
|---|---|---|---|
| No notation rolls yet. | |||
The muted dice shapes are only an empty-state cue. No random face is selected until you press Roll dice. After a roll, the stage shows the actual generated dice and the total, range, and recent history update.
Standard d4 through d100 choices stay one tap away, while sound is secondary so the roll itself remains the obvious first action.
Use the product first, then apply these tool-specific checks to verify assumptions, interpret the result, and hand it off safely without moving the primary workflow below generic content.
Compact dice notation is most useful when another person can reproduce the rule before the roll. Write the count, sides, keep rule, and flat modifier explicitly, such as 2d20kh1+5, rather than describing the rule after seeing a result. The notation panel is designed for tabletop, classroom, and probability exercises; it does not change the underlying random draw after the expression is submitted.
Advantage-style keep-highest and keep-lowest rules change which dice contribute to the total. Decide the keep count before rolling and preserve the individual values in the history so the discarded values remain visible. Editing a rule after seeing the roll creates a different experiment and should be recorded as a new roll rather than treated as the same event.
For an NdS roll every raw die must be an integer from 1 through S. The enhanced notation workflow uses browser cryptographic randomness when the platform exposes it and records the source in the report; otherwise it falls back to the browser pseudorandom generator. Randomness does not guarantee a balanced short session, so streaks are possible even when every draw is valid.
History, average, minimum, and maximum summarize rolls already generated. They do not predict the next result or make future outcomes more likely to compensate for earlier ones. For a classroom probability check, compare a sufficiently large sample with the theoretical range and expected mean, and keep the raw roll history if another person needs to audit the calculation.
WebToolArc