Seeded dice roller

That makes random-looking rolls practical for tests, game replays, classroom exercises, and shared scenarios where every participant needs identical results. Enter how many dice to roll, choose the number of sides on each die, and provide any non-empty text seed. The result lists each individual roll in order and adds them into one total.

Make a dice roll reproducible

Ordinary dice tools are designed to surprise you, but reproducibility matters whenever another person or another run must see exactly the same outcome. This seeded dice roller converts your seed text into a fixed internal starting value, then advances a deterministic generator once for every accepted die result. If you submit the same number of dice, the same number of sides, and the same seed, the ordered rolls and total remain identical. Changing even one of those inputs creates a different scenario. A seed can be a test case name, a match identifier, a lesson label, or any other non-empty text that your team can store and share. Treat the seed as part of the scenario definition rather than as a secret. The response includes a rolls array so you can inspect each die, not merely the sum, which is useful for rules involving doubles, highest or lowest dice, individual successes, or later replay. It also includes the total so common checks need no extra arithmetic.

Choose dice and sides carefully

The dice field controls how many independent results are returned, while sides defines the inclusive range of every result. For example, six-sided dice always produce integers from 1 through 6, and twenty-sided dice always produce integers from 1 through 20. The capability accepts between 1 and 1,000 dice and between 2 and 1,000,000 sides per die. A die with fewer than two sides is rejected because it does not represent a meaningful random choice, and fractional or non-numeric values are rejected rather than silently rounded. The implementation maps generated 32-bit values to die faces with rejection sampling. That avoids the small modulo bias created when the generator range is not evenly divisible by the number of sides. Results are deterministic across server and browser execution because the algorithm uses only explicitly defined integer operations. It does not call a system random source, inspect the clock, depend on locale, or contact another service. These properties make the response suitable for fixtures and repeatable demonstrations, although it is not intended for cryptography, wagering, or security-sensitive lotteries.

Store and replay complete scenarios

For reliable replay, save all three inputs together: dice, sides, and seed. Saving only the total is not enough when game logic later needs the individual rolls, and saving only the seed is not enough when the dice configuration can change. A test suite can derive a seed from a case name, call the capability, and keep the returned rolls as readable evidence. A game master can publish a scenario seed before a session so participants can reproduce an encounter without exposing results in advance. Teachers can give every student the same generated data while still presenting an exercise that looks like a natural dice experiment. The sequence is tied to this capability's documented algorithm; because the capability begins in beta, applications that require permanent archival replay should also store the returned rolls. API automation costs $0.002 per request, while the browser runner can perform the same pure computation locally. When comparing two runs, compare the ordered rolls array as well as the total: different sequences can share a sum, so the array is the stronger proof that the entire scenario was reproduced.