A terminal-based implementation of Scout (aka SCOUT!), the award-winning card game designed by Kei Kajino.

Play against AI opponents right in your terminal with a colorful TUI powered by Bubble Tea.

Menu	In-Game

Quick install (macOS / Linux):

curl -sL https://raw.githubusercontent.com/daghlny/scout_cli/main/install.sh | sh

With Go:

go install github.com/daghlny/scout_cli@latest

Manual download: grab the binary for your platform from GitHub Releases.

Then run:

scout_cli

• 3-5 player games (you vs AI bots)
• Full Scout rules: double-sided cards, Show / Scout / Scout & Show actions
• Smart AI with tactical awareness and defensive play
• LLM AI mode: works with any OpenAI-compatible API (OpenAI, DeepSeek, Ollama, etc.)
• Colorful card rendering in terminal
• Multi-round scoring with final leaderboard

Default (Smart AI):

scout_cli

LLM AI — each bot calls a large language model to decide its move:

export OPENAI_API_KEY=your_key
scout_cli -ai=llm

Works with any OpenAI-compatible API. Configure via environment variables:

Examples:

# OpenAI
OPENAI_API_KEY=sk-xxx scout_cli -ai=llm

# DeepSeek
OPENAI_BASE_URL=https://api.deepseek.com/v1 OPENAI_API_KEY=sk-xxx OPENAI_MODEL=deepseek-chat scout_cli -ai=llm

# Local Ollama
OPENAI_BASE_URL=http://localhost:11434/v1 OPENAI_API_KEY=ollama OPENAI_MODEL=llama3 scout_cli -ai=llm

The LLM receives game rules, current hand, table state, and action history, then reasons through function calling. Falls back to Smart AI automatically on API errors.

Scout uses 45 unique double-sided cards (each with two different numbers 1-10). Cards in your hand cannot be rearranged — you can only insert new cards via scouting. On your turn, choose one action:

• Show: Play adjacent cards from your hand that beat the table combo
• Scout: Take a card from either end of the table combo and insert it into your hand
• Scout & Show: Do both (once per round)

For full rules, see: https://boardgamegeek.com/boardgame/291453/scout

Scout（马戏星探）的终端命令行版本，基于 Bubble Tea 构建。

主菜单	游戏进行中

一键安装（macOS / Linux，无需 Go 环境）：

curl -sL https://raw.githubusercontent.com/daghlny/scout_cli/main/install.sh | sh

有 Go 环境：

go install github.com/daghlny/scout_cli@latest

手动下载：从 GitHub Releases 下载对应平台的二进制文件（支持 macOS / Linux / Windows）。

运行：

scout_cli

• 支持 3-5 人游戏（你 vs AI 对手）
• 完整的 Scout 规则：双面牌、出牌 / 招募 / 双重行动
• 具备攻防意识的智能 AI，懂得保留强牌与战术防守
• LLM AI 模式：支持任意 OpenAI 兼容接口（OpenAI、DeepSeek、Ollama 等）
• 终端内彩色扑克牌渲染
• 多轮计分与最终排行榜

默认（Smart AI）：

scout_cli

LLM AI — 每个 bot 调用大语言模型来决策：

export OPENAI_API_KEY=your_key
scout_cli -ai=llm

支持任意 OpenAI 兼容接口，通过环境变量配置：

示例：

# OpenAI
OPENAI_API_KEY=sk-xxx scout_cli -ai=llm

# DeepSeek
OPENAI_BASE_URL=https://api.deepseek.com/v1 OPENAI_API_KEY=sk-xxx OPENAI_MODEL=deepseek-chat scout_cli -ai=llm

# 本地 Ollama
OPENAI_BASE_URL=http://localhost:11434/v1 OPENAI_API_KEY=ollama OPENAI_MODEL=llama3 scout_cli -ai=llm

LLM 会接收游戏规则、当前手牌、桌面状态和历史行动记录，通过 function calling 进行结构化推理。API 出错时自动回退到 Smart AI。

Scout 使用 45 张双面牌（每张牌正反两面各有一个 1-10 的数字）。手牌不可重新排列，只能通过招募插入新牌。每回合选择一个行动：

• 出牌（Show）：打出手中相邻的牌组合，必须强过桌面上的牌
• 招募（Scout）：从桌面牌组的左端或右端取一张牌插入手中
• 双重行动（Scout & Show）：先招募再出牌（每轮限一次）

完整规则参见：https://boardgamegeek.com/boardgame/291453/scout