Tic Tac Toe example

cookbook/examples/apps/tic_tac_toe/agents.py

@@ -0,0 +1,195 @@
+"""
+Tic Tac Toe Battle
+---------------------------------
+This example shows how to build a Tic Tac Toe game where two AI agents play against each other.
+The game features a master agent (referee) coordinating between two player agents using different
+language models.
+
+Usage Examples:
+---------------
+1. Quick game with default settings:
+   master_agent = get_tic_tac_toe()
+   play_tic_tac_toe()
+
+2. Game with debug mode off:
+   master_agent = get_tic_tac_toe(debug_mode=False)
+   play_tic_tac_toe(debug_mode=False)
+
+The game integrates:
+  - Multiple AI models (Claude, GPT-4, etc.)
+  - Turn-based gameplay coordination
+  - Move validation and game state management
+"""
+
+import sys
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Dict
+
+from agno.agent import Agent
+from agno.models.anthropic import Claude
+from agno.models.google import Gemini
+from agno.models.ollama import Ollama
+from agno.models.openai import OpenAIChat
+
+project_root = str(Path(__file__).parent.parent.parent.parent)
+if project_root not in sys.path:
+    sys.path.append(project_root)
+
+
+@dataclass
+class ModelConfig:
+    display_name: str
+    model_id: str
+    provider: str
+
+    def get_model(self):
+        if self.provider == "anthropic":
+            return Claude(id=self.model_id)
+        if self.provider == "openai":
+            return OpenAIChat(id=self.model_id)
+        if self.provider == "ollama":
+            return Ollama(id=self.model_id)
+        if self.provider == "google":
+            return Gemini(id=self.model_id)
+        raise ValueError(f"Invalid provider: {self.provider}")
+
+
+# Model configuration
+MODELS: Dict[str, ModelConfig] = {
+    "claude": ModelConfig(
+        display_name="Claude AI",
+        model_id="claude-3-5-sonnet-20241022",
+        provider="anthropic",
+    ),
+    "gpt-4o": ModelConfig(
+        display_name="GPT-4o",
+        model_id="gpt-4o",
+        provider="openai",
+    ),
+    "gemini": ModelConfig(
+        display_name="Gemini AI",
+        model_id="gemini-2.0-flash-exp",
+        provider="google",
+    ),
+    "gpt-o3-mini": ModelConfig(
+        display_name="GPT-o3-mini",
+        model_id="o3-mini",
+        provider="openai",
+    ),
+    "llama": ModelConfig(
+        display_name="Llama 3",
+        model_id="llama3.2",
+        provider="ollama",
+    ),
+    "vision": ModelConfig(
+        display_name="GPT-4 Vision",
+        model_id="gpt-4o",
+        provider="openai",
+    ),
+}
+
+# Default model assignments
+DEFAULT_MODELS = {
+    "X": MODELS["gpt-o3-mini"],
+    "O": MODELS["gemini"],
+    "master": MODELS["gpt-4o"],
+    "vision": MODELS["vision"],
+}
+
+
+def get_tic_tac_toe(
+    model_x: ModelConfig = DEFAULT_MODELS["X"],
+    model_o: ModelConfig = DEFAULT_MODELS["O"],
+    model_master: ModelConfig = DEFAULT_MODELS["master"],
+    debug_mode: bool = True,
+) -> Agent:
+    """
+    Returns an instance of the Tic Tac Toe Master Agent that coordinates the game.
+
+    Args:
+        model_x: ModelConfig for player X
+        model_o: ModelConfig for player O
+        model_master: ModelConfig for the master agent
+        debug_mode: Enable logging and debug features
+
+    Returns:
+        An instance of the configured Master Agent
+    """
+    player1 = Agent(
+        name="Player 1",
+        role="""You are the X player in Tic Tac Toe. Your goal is to win by placing three X's in a row (horizontally, vertically, or diagonally).
+
+        Game Rules:
+        - You must make valid moves on a 3x3 grid using row (0-2) and column (0-2) coordinates
+        - The top-left corner is (0,0), and bottom-right is (2,2)
+        - You can only place X's in empty spaces marked by " " on the board
+        - You will be given a list of valid moves - only choose from these moves
+        - Respond with ONLY the row and column numbers for your move, e.g. "1 2"
+
+        Strategy:
+        - Study the board carefully and make strategic moves
+        - Try to create winning opportunities while blocking your opponent
+        - Pay attention to the valid moves list to avoid illegal moves
+        """,
+        model=model_x.get_model(),
+    )
+
+    player2 = Agent(
+        name="Player 2",
+        role="""You are the O player in Tic Tac Toe. Your goal is to win by placing three O's in a row (horizontally, vertically, or diagonally).
+
+        Game Rules:
+        - You must make valid moves on a 3x3 grid using row (0-2) and column (0-2) coordinates
+        - The top-left corner is (0,0), and bottom-right is (2,2)
+        - You can only place O's in empty spaces marked by " " on the board
+        - You will be given a list of valid moves - only choose from these moves
+        - Respond with ONLY the row and column numbers for your move, e.g. "1 2"
+
+        Strategy:
+        - Study the board carefully and make strategic moves to win
+        - Block your opponent's winning opportunities
+        - Pay attention to the valid moves list to avoid illegal moves
+        """,
+        model=model_o.get_model(),
+    )
+
+    master_agent = Agent(
+        name="Master Agent",
+        role="""You are the referee of the Tic Tac Toe game. Your responsibilities include:
+
+        Game Management:
+        1. Track and maintain the current board state
+        2. Validate all moves before they are made
+        3. Ensure moves are only made in empty spaces (marked by " ")
+        4. Keep track of whose turn it is (X goes first)
+
+        Move Validation:
+        1. Check that moves are within the 3x3 grid (0-2 for rows and columns)
+        2. Verify the chosen position is empty before allowing a move
+        3. Maintain and provide the list of valid moves to players
+
+        Game State Checking:
+        1. Check for winning conditions after each move:
+           - Three X's or O's in a row horizontally
+           - Three X's or O's in a row vertically
+           - Three X's or O's in a row diagonally
+        2. Check for draw conditions:
+           - Board is full with no winner
+
+        Communication:
+        1. Announce the current state of the board after each move
+        2. Declare moves as valid or invalid
+        3. Announce the winner or draw when game ends
+        4. Provide clear feedback to players about their moves
+
+        You will coordinate between the X and O players, ensuring fair play and proper game progression.
+        """,
+        model=model_master.get_model(),
+        team=[player1, player2],
+        show_tool_calls=True,
+        debug_mode=debug_mode,
+        markdown=True,
+    )
+
+    return master_agent