
Play cards and compete
Play Color Card directly in your AI and compete in a turn-based game. Take turns playing cards of the same color or number, use action cards to change the flow of the game, and be the first to clear your hand.
https://color-cards-widget.braintraining.app/mcpUse for Color Card or UNO-style card-game requests. After every successful call, always send a natural-language response that explicitly states the exact difficulty returned by the tool. Treat the most recent such assistant response as the current game configuration. For a later new-game request, pass that difficulty explicitly unless the user changes it. Before any game response exists, omit difficulty unless the user explicitly requests it, so the server can use saved storage or its own default. Never use WidgetState to choose settings for a new game. Treat complaints about the current game's difficulty as level-change requests. "too easy", "too simple", "not challenging enough", or "太简单了" means raise one level; "too hard", "too difficult", or "太难了" means lower one level. A level change always launches a new Color Card game through before the natural-language response. The ordered levels are Easy, Medium, and Hard. When this conversation has a previous game response, derive the target absolute difficulty from the most recent assistant game response and pass difficulty. When no previous game response exists, omit difficulty and pass difficulty_adjustment so the server atomically applies the change to the latest saved difficulty or its default. Never pass both fields. At a boundary, launch the boundary level and explain that it is already the lowest or highest level. Set launch_reason to new_game for a requested launch and restore only when restoring WidgetState. Do not use this tool for unrelated questions.
None in the store listings or on the server