Teams 2.0

README.md

@@ -190,6 +190,7 @@ from agno.agent import Agent
 from agno.models.openai import OpenAIChat
 from agno.tools.duckduckgo import DuckDuckGoTools
 from agno.tools.yfinance import YFinanceTools
+from agno.team.team import Team
 
 web_agent = Agent(
     name="Web Agent",
@@ -212,8 +213,10 @@ finance_agent = Agent(
 )
 
 agent_team = Agent(
-    team=[web_agent, finance_agent],
+    mode="coordinator",
+    members=[web_agent, finance_agent],
     model=OpenAIChat(id="gpt-4o"),
+    success_criteria="A comprehensive financial news report with clear sections and data-driven insights.",
     instructions=["Always include sources", "Use tables to display data"],
     show_tool_calls=True,
     markdown=True,

cookbook/README.md

@@ -17,7 +17,7 @@ The concepts cookbook walks through the core concepts of Agno.
 - [Reasoning](./agent_concepts/reasoning)
 - [Vector DBs](./agent_concepts/vector_dbs)
 - [Multi-modal Agents](./agent_concepts/multimodal)
-- [Agent Teams](./agent_concepts/teams)
+- [Agent Teams](teams)
 - [Hybrid Search](./agent_concepts/hybrid_search)
 - [Agent Session](./agent_concepts/agent_session)
 - [Other](./agent_concepts/other)

cookbook/models/xai/.gitignore → cookbook/agent_concepts/.gitignore

@@ -3,4 +3,4 @@
 *.mp3
 *.wav
 *.mp4
-*.mp3
+*.mp3

cookbook/examples/apps/chess_team/agents.py

@@ -8,10 +8,10 @@
 Usage Examples:
 ---------------
 1. Quick game with default settings:
-   agents = get_chess_teams()
+   team = get_chess_team()
 
 2. Game with debug mode off:
-   agents = get_chess_teams(debug_mode=False)
+   team = get_chess_team(debug_mode=False)
 
 The game integrates:
   - Multiple AI models (Claude, GPT-4, etc.)
@@ -22,13 +22,13 @@
 
 import sys
 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.groq import Groq
 from agno.models.openai import OpenAIChat
+from agno.team.team import Team
 from agno.utils.log import logger
 
 project_root = str(Path(__file__).parent.parent.parent.parent)
@@ -66,6 +66,7 @@ def get_model_for_provider(provider: str, model_name: str):
             return Claude(
                 id="claude-3-7-sonnet-20250219",
                 max_tokens=8192,
+                thinking={"type": "enabled", "budget_tokens": 4096},
             )
         else:
             return Claude(id=model_name)
@@ -75,14 +76,14 @@ def get_model_for_provider(provider: str, model_name: str):
         raise ValueError(f"Unsupported model provider: {provider}")
 
 
-def get_chess_teams(
+def get_chess_team(
     white_model: str = "openai:gpt-4o",
     black_model: str = "anthropic:claude-3-7-sonnet",
     master_model: str = "openai:gpt-4o",
     debug_mode: bool = True,
-) -> Dict[str, Agent]:
+) -> Team:
     """
-    Returns a dictionary of chess agents with specific roles.
+    Returns a chess team with specialized agents for white pieces, black pieces, and game master.
 
     Args:
         white_model: Model for white piece strategy
@@ -91,22 +92,20 @@ def get_chess_teams(
         debug_mode: Enable logging and debug features
 
     Returns:
-        Dictionary of configured agents
+        Team instance configured for chess gameplay
     """
     try:
-        # Parse model providers and names
         white_provider, white_name = white_model.split(":")
         black_provider, black_name = black_model.split(":")
         master_provider, master_name = master_model.split(":")
 
-        # Create model instances
         white_piece_model = get_model_for_provider(white_provider, white_name)
         black_piece_model = get_model_for_provider(black_provider, black_name)
         master_model = get_model_for_provider(master_provider, master_name)
 
-        # Create agents
         white_piece_agent = Agent(
             name="white_piece_agent",
+            role="White Piece Strategist",
             description="""You are a chess strategist for white pieces. Given a list of legal moves,
                     analyze them and choose the best one based on standard chess strategy.
                     Consider piece development, center control, and king safety.
@@ -117,6 +116,7 @@ def get_chess_teams(
 
         black_piece_agent = Agent(
             name="black_piece_agent",
+            role="Black Piece Strategist",
             description="""You are a chess strategist for black pieces. Given a list of legal moves,
                     analyze them and choose the best one based on standard chess strategy.
                     Consider piece development, center control, and king safety.
@@ -125,31 +125,39 @@ def get_chess_teams(
             debug_mode=debug_mode,
         )
 
-        master_agent = Agent(
-            name="master_agent",
-            description="""You are a chess master overseeing the game. Your responsibilities:
-                    1. Analyze the current board state to determine if the game has ended
-                    2. Check for checkmate, stalemate, draw by repetition, or insufficient material
-                    3. Provide commentary on the current state of the game
-                    4. Evaluate the position and suggest who has an advantage
-
-                    Respond with a JSON object containing:
-                    {
-                        "game_over": true/false,
-                        "result": "white_win"/"black_win"/"draw"/null,
-                        "reason": "explanation if game is over",
-                        "commentary": "brief analysis of the position",
-                        "advantage": "white"/"black"/"equal"
-                    }""",
+        return Team(
+            name="Chess Team",
+            mode="router",
             model=master_model,
+            success_criteria="The game is completed with a win, loss, or draw",
+            members=[white_piece_agent, black_piece_agent],
+            instructions=[
+                "You are the chess game coordinator and master analyst.",
+                "Your role is to coordinate between two player agents and provide game analysis:",
+                "1. white_piece_agent - Makes moves for white pieces",
+                "2. black_piece_agent - Makes moves for black pieces",
+                "",
+                "When receiving a task:",
+                "1. Check the 'current_player' in the context",
+                "2. If current_player is white_piece_agent or black_piece_agent:",
+                "   - Forward the move request to that agent",
+                "   - Return their move response directly without modification",
+                "3. If no current_player is specified:",
+                "   - This indicates a request for position analysis",
+                "   - Analyze the position yourself and respond with a JSON object:",
+                "   {",
+                "       'game_over': true/false,",
+                "       'result': 'white_win'/'black_win'/'draw'/null,",
+                "   }",
+                "",
+                "Do not modify player agent responses.",
+                "For analysis requests, provide detailed evaluation of the position.",
+            ],
             debug_mode=debug_mode,
+            markdown=True,
+            show_members_responses=True,
         )
 
-        return {
-            "white_piece_agent": white_piece_agent,
-            "black_piece_agent": black_piece_agent,
-            "master_agent": master_agent,
-        }
     except Exception as e:
-        logger.error(f"Error initializing agents: {str(e)}")
+        logger.error(f"Error initializing chess team: {str(e)}")
         raise