Cribbage Game

A complete two-player Cribbage hand

This specialist models the full rhythm of two-player Cribbage rather than reducing the game to a hand-scoring quiz. Six cards are dealt to each player. Each side chooses two cards for the dealer’s crib, the starter card is cut, pegging is played without allowing the running total to pass thirty-one, and the show is counted in the traditional order: non-dealer hand, dealer hand, then dealer crib. Race to 121 carries those points across hands while Practice Hand and Daily Cribbage keep the session compact.

The crib changes discard strategy

You are not simply throwing away the two cards you like least. The dealer owns the four-card crib, so two connected cards can be valuable gifts to your own future count and dangerous gifts to the opponent. The local AI uses a deterministic discard heuristic that considers pairs, suited neighbors, nearby ranks and fifteen-making relationships, with the sign of that crib value changing according to who deals. It is a study heuristic, not a claim of tournament-level discard analysis.

The starter can score before pegging begins

After both players contribute to the crib, the top remaining card becomes the starter. If that starter is a Jack, the dealer immediately scores two points for his heels. The same starter is later combined with each four-card hand and with the crib during the show. That single public card can therefore change flushes, fifteens, runs and nobs even before the players begin pegging.

Pegging never allows the count above thirty-one

Players alternate one card at a time and add the card’s Cribbage value to a running count. Aces are one, number cards use face value and face cards are ten. A card that would push the running total above thirty-one is not legal. If you cannot play any card, you say Go. The opponent may continue while a legal card remains, then the sequence resets and the player who did not place the final card leads the next count.

Fifteen and thirty-one each score two

During pegging, reaching exactly fifteen scores two points and reaching exactly thirty-one also scores two. Thirty-one closes the current sequence immediately. The app calculates those points directly from the public sequence and running total, so the visual score track and the verifier use the same rule function.

Pairs and runs depend on the newest cards

Two equal ranks played consecutively score two points. Three equal ranks score six, and four equal ranks score twelve. Pegging runs are based on the newest consecutive group of cards and do not require the cards to have been played in numerical order. For example, a public sequence ending five, three, four contains a three-card run. The engine searches the longest valid suffix rather than looking only at the last two ranks.

Go and last-card scoring are explicit

If a player has no legal card at the current count, Go is a real action rather than a hidden automatic skip. The opponent can keep playing while possible. When the sequence ends below thirty-one, the player who placed the final card scores one for last card or Go. A thirty-one already scores its own two points and does not receive another last-card point.

The show counts every fifteen combination

Once pegging is finished, each four-card hand is combined with the starter. Every distinct subset of cards that totals fifteen scores two points. That means one five-card group can contain several different fifteen combinations. The scorer enumerates those combinations directly, so unusual multi-fifteen hands do not depend on a simplified pattern table.

Pairs and duplicated runs can stack

Every pair of equal ranks is worth two during the show. Runs score their length, but duplicate ranks can create multiple distinct runs. A double run therefore needs more than a simple “are these ranks consecutive?” check. The scorer evaluates valid card subsets, identifies the longest run length present, and counts every distinct run of that maximum length.

Flush rules differ between a hand and the crib

In an ordinary hand, four cards of one suit score four even if the starter has another suit; a matching starter increases the flush to five. The crib is stricter: a crib flush scores only when all four crib cards and the starter have the same suit. This small exception is covered by independent fixtures because it is easy to implement incorrectly when one general flush rule is reused everywhere.

Nobs is separate from his heels

Nobs scores one point when the four-card hand contains a Jack matching the starter’s suit. His heels is the different two-point event where the starter itself is a Jack and the dealer scores before pegging. The interface and core keep these rules separate so the famous terminology maps cleanly to the correct scoring stage.

The maximum hand score is twenty-nine

The verifier includes the traditional perfect twenty-nine-point hand. Three fives in the hand, a Jack matching the starter suit, and the fourth five as starter combine the maximum number of fifteens and pairs with the one nobs point. Testing that exact total is a compact way to exercise several show-scoring rules at once.

The computer has no access to your hidden hand

The AI boundary is intentionally smaller than the full game state. During pegging it receives its own remaining cards, the public starter, public sequence, running count, dealer information and scores. It does not receive your hidden cards or unseen deck order. Crib-discard decisions receive only the AI’s six cards and whether the AI owns the crib. That keeps difficulty changes inside a fair local information model.

Easy, Standard and Hard share one legality engine

Every difficulty level must choose from the same legal cards and obey the same thirty-one boundary. Easy preserves more seeded variation. Standard rewards immediate scoring while trying to avoid awkward totals. Hard adds a little more weight to public sequence shape and the flexibility of the cards it keeps. These labels describe local heuristics only; the product does not present them as expert, championship or statistically rated play.

Practice seeds make deals reproducible

Opening Practice without a seed generates one and stabilizes it in the URL. The deal and deterministic AI tie-breaks then belong to that seed. Reopening the same seed restores the same autosave if one exists. Fresh Hand creates a different seed. Your discard and pegging choices still determine the path, so deterministic means reproducible starting conditions rather than a scripted outcome.

Hint and Undo are visible assistance

Hint can identify a pair of candidate crib discards or suggest a legal pegging card. Undo returns to the saved state before a user decision. Either action changes the hand to Assisted. The distinction allows the tools to teach scoring and card management without presenting a coached result as an unaided one.

Local records remain factual

The browser stores hands played, hand wins, clean wins, game wins, best show score, Daily history and Daily streak. There is no external multiplayer ladder behind the page, so it does not manufacture an Elo, percentile, win probability or worldwide rank. The records are limited to facts the local game can actually observe.

Strategy and arithmetic, with no wagering mechanics

Cribbage uses a standard deck here only as a free strategy and counting game. There are no wagers, chips, prizes, stakes, purchases or paid random mechanics. The repeat value comes from choosing the crib discard, reading the public pegging sequence and learning the exact arithmetic of the show.