Preparing the forest board…
These English boards use a small, authored vocabulary and directed graph. Scores and connections are illustrative game fixtures, not human association measurements or results from a three-million-word corpus.
Lexitriad ranks supported answers within each semantic type. Two optional scouts reveal candidate ranks without submitting them; six guesses lock the three slots. Results disclose scout use. Low-support entries and words outside this demo do not use an attempt. Unknown does not mean your association is wrong.
Sense Seals accepts only published demo edges. Each costs one hop. Its optimum is computed after applying the ordered passport rules. Building from the arrival side checks incoming edges.
Plural aliases share a word identity. Types are word-level labels; sense switching is allowed. Both endpoints are excluded from stamps, and a word occurrence earns at most one stamp.
Lexitriad and Sense Seals use the frozen demo-en-v1 dataset. Replykin prompt sets and Word Roast challenges retain their own authored edition IDs. Lexitriad and Sense Seals progress is saved in this browser and checked again after reload. Clearing browser storage removes it; failed saving is shown beside the game. Unfinished Replykin and Word Roast answer drafts stay in this browser; clearing browser storage removes them. Completed Replykin challenges and Word Roast duels are saved on the server. An anonymous browser cookie remembers your challenge ownership and votes. Clearing cookies creates a new identity. Scores are practice results, with no competitive leaderboard.