Merge pull request #108 from brnaba-aws/main

Split MAO python package dependency into feature (anthropic and openai) and removed Bedrock dependency in Classifier and Orchestrator

README.md

@@ -69,8 +69,8 @@ To quickly get a feel for the Multi-Agent Orchestrator, check out our [Demo App]
 Get hands-on experience with the Multi-Agent Orchestrator through our diverse set of examples:
 
 - **Ready-to-run Scripts**: Start locally with our collection of standalone scripts in both Python and TypeScript.
-- **Demo Applications**: 
-  - [Chat Demo App](https://awslabs.github.io/multi-agent-orchestrator/cookbook/examples/chat-demo-app/): 
+- **Demo Applications**:
+  - [Chat Demo App](https://awslabs.github.io/multi-agent-orchestrator/cookbook/examples/chat-demo-app/):
     - Explore multiple specialized agents handling various domains like travel, weather, math, and health
   - [E-commerce Support Simulator](https://awslabs.github.io/multi-agent-orchestrator/cookbook/examples/ecommerce-support-simulator/): Experience AI-powered customer support with:
     - Automated response generation for common queries
@@ -83,8 +83,8 @@ Get hands-on experience with the Multi-Agent Orchestrator through our diverse se
   - [`chat-chainlit-app`](https://github.com/awslabs/multi-agent-orchestrator/tree/main/examples/chat-chainlit-app): Chat application built with Chainlit
   - [`fast-api-streaming`](https://github.com/awslabs/multi-agent-orchestrator/tree/main/examples/fast-api-streaming): FastAPI implementation with streaming support
   - [`text-2-structured-output`](https://github.com/awslabs/multi-agent-orchestrator/tree/main/examples/text-2-structured-output): Natural Language to Structured Data
-  
-  
+
+
 All examples are available in both Python and TypeScript implementations. Check out our [documentation](https://awslabs.github.io/multi-agent-orchestrator/) for comprehensive guides on setting up and using the Multi-Agent Orchestrator!
 
 
@@ -192,7 +192,7 @@ if (response.streaming == true) {
 
 ### Python Version
 
-#### Installation
+#### Core Installation
 
 ```bash
 # Optional: Set up a virtual environment
@@ -201,7 +201,7 @@ source venv/bin/activate  # On Windows use `venv\Scripts\activate`
 pip install multi-agent-orchestrator
 ```
 
-#### Usage
+#### Default Usage
 
 Here's an equivalent Python example demonstrating the use of the Multi-Agent Orchestrator with a Bedrock LLM Agent and a Lex Bot Agent:
 
@@ -289,6 +289,19 @@ These examples showcase:
 3. The orchestrator's ability to route requests to the most appropriate agent based on the input.
 4. Handling of both streaming and non-streaming responses from different types of agents.
 
+### Working with Anthropic or OpenAI
+If you want to use Anthropic or OpenAI for classifier and/or agents, make sure to install the multi-agent-orchestrator with the relevant extra feature.
+```bash
+pip install multi-agent-orchestrator[anthropic]
+pip install multi-agent-orchestrator[openai]
+```
+
+### Full package installation
+For a complete installation (including Anthropic and OpenAi):
+```bash
+pip install multi-agent-orchestrator[all]
+```
+
 
 ## 🤝 Contributing
 

docs/src/content/docs/classifiers/built-in/anthropic-classifier.mdx

@@ -16,6 +16,11 @@ The Anthropic Classifier extends the abstract `Classifier` class and uses the An
 
 ### Basic Usage
 
+⚠️ To use Anthropic Classifier, make sure you have installed the multi-agent-orchestrator with anthropic feature (see [python installation](/multi-agent-orchestrator/general/quickstart#-get-started))
+```bash
+pip install multi-agent-orchestrator[anthropic]
+```
+
 To use the AnthropicClassifier, you need to create an instance with your Anthropic API key and pass it to the Multi-Agent Orchestrator:
 
 import { Tabs, TabItem } from '@astrojs/starlight/components';

docs/src/content/docs/classifiers/built-in/openai-classifier.mdx

@@ -16,6 +16,12 @@ The OpenAI Classifier extends the abstract `Classifier` class and uses the OpenA
 
 ## Basic Usage
 
+⚠️ To use OpenAI Classifier, make sure you have installed the multi-agent-orchestrator with openai feature (see [python installation](/multi-agent-orchestrator/general/quickstart#-get-started))
+
+```bash
+pip install multi-agent-orchestrator[openai]
+```
+
 To use the OpenAIClassifier, you need to create an instance with your OpenAI API key and pass it to the Multi-Agent Orchestrator:
 
 import { Tabs, TabItem } from '@astrojs/starlight/components';

docs/src/content/docs/classifiers/overview.mdx

@@ -11,6 +11,9 @@ The Classifier is a crucial component of the Multi-Agent Orchestrator, responsib
 
 - **[Anthropic Classifier](/multi-agent-orchestrator/classifiers/built-in/anthropic-classifier)** uses Anthropic's AI models for intent classification. It provides an alternative option for users who prefer or have access to Anthropic's services.
 
+- **[OpenAI Classifier](/multi-agent-orchestrator/classifiers/built-in/openai-classifier)** uses OpenAI's AI models for intent classification. It provides an alternative option for users who prefer or have access to OpenAI's services.
+
+
 ### Process Flow
 
 Regardless of the classifier used, the general process remains the same:
@@ -23,11 +26,11 @@ Regardless of the classifier used, the general process remains the same:
 
 By default, if no agent is selected the orchestrator can be configured to:
 
-A. Use a default agent (a **[BedrockLLMAgent](/multi-agent-orchestrator/agents/built-in/bedrock-llm-agent)**)
+A. Use a default agent (a **[BedrockLLMAgent for example](/multi-agent-orchestrator/agents/built-in/bedrock-llm-agent)**)
 
 B. Return a message prompting the user for more information.
 
-This behavior can be customized using the `USE_DEFAULT_AGENT_IF_NONE_IDENTIFIED` and `NO_SELECTED_AGENT_MESSAGE` configuration options in the orchestrator. 
+This behavior can be customized using the `USE_DEFAULT_AGENT_IF_NONE_IDENTIFIED` and `NO_SELECTED_AGENT_MESSAGE` configuration options in the orchestrator.
 
 For a detailed explanation of these options and other orchestrator configurations, please refer to the [Orchestrator Overview](/multi-agent-orchestrator/orchestrator/overview#agent-selection-and-default-behavior) page.
 
@@ -105,13 +108,13 @@ You can test any Classifier directly using the `classify` method:
         user_input = "What are the symptoms of the flu?"
         user_id = "test_user"
         session_id = "test_session"
-        
+
         # Fetch chat history (this might vary depending on your implementation)
         chat_history = await orchestrator.storage.fetch_all_chats(user_id, session_id)
-        
+
         # Classify the input
         response = await orchestrator.classifier.classify(user_input, chat_history)
-        
+
         print('\n** RESPONSE ** \n')
         print(f" > Agent ID: {response.selected_agent.id if response.selected_agent else 'None'}")
         print(f" > Agent Name: {response.selected_agent.name if response.selected_agent else 'None'}")

docs/src/content/docs/general/quickstart.mdx

@@ -38,7 +38,7 @@ To help you kickstart with the Multi-Agent Orchestrator framework, we'll walk yo
 
 2. Authenticate with your AWS account
 
-This quickstart demonstrates the use of Amazon Bedrock for both classification and agent responses. 
+This quickstart demonstrates the use of Amazon Bedrock for both classification and agent responses.
 
 To authenticate with your AWS account, follow these steps:
 
@@ -61,8 +61,8 @@ By default, the framework is configured as follows:
 <br/>
 
 > **Important**
-> 
->   These are merely default settings and can be easily changed to suit your needs or preferences. 
+>
+>   These are merely default settings and can be easily changed to suit your needs or preferences.
 
 <br/>
 
@@ -91,7 +91,10 @@ Ensure you have [requested access](https://docs.aws.amazon.com/bedrock/latest/us
   </TabItem>
   <TabItem label="Python" icon="seti:python">
     ```bash
-    pip install multi-agent-orchestrator
+    pip install multi-agent-orchestrator # for core dependencies
+    pip install multi-agent-orchestrator[anthropic] # for Anthropic classifier and agent
+    pip install multi-agent-orchestrator[openai] # for OpenAI classifier and agent
+    pip install multi-agent-orchestrator[all] # for all packages including Anthropic and OpenAI
     ```
   </TabItem>
 </Tabs>
@@ -165,7 +168,7 @@ Ensure you have [requested access](https://docs.aws.amazon.com/bedrock/latest/us
         streaming: true
       })
     );
-    
+
     orchestrator.addAgent(
       new BedrockLLMAgent({
         name: "Health Agent",
@@ -227,7 +230,7 @@ Ensure you have [requested access](https://docs.aws.amazon.com/bedrock/latest/us
             console.error("Received unexpected chunk type:", typeof chunk);
           }
         }
-        console.log(); 
+        console.log();
       } catch (error) {
         console.error("An error occurred:", error);
         // Here you could also add more specific error handling if needed

docs/src/content/docs/orchestrator/overview.mdx

@@ -49,6 +49,7 @@ The Orchestrator accepts an `OrchestratorConfig` object during initialization. A
    - `GENERAL_ROUTING_ERROR_MSG_MESSAGE`: Custom message for general routing errors.
 3. `logger`: Custom logger instance. If not provided, a default logger will be used.
 4. `classifier`: Custom classifier instance. If not provided, a `BedrockClassifier` will be used.
+5. `default_agent`: A default agent when the classifier could not determine the most suitable agent.
 
 ### Example with all options
 
@@ -107,7 +108,12 @@ Here's an example that demonstrates how to initialize the Orchestrator with all
       ),
       storage=DynamoDBChatStorage(),
       classifier=CustomClassifier(),
-      logger=CustomLogger()
+      logger=CustomLogger(),
+      default_agent=BedrockLLMAgent(BedrockLLMAgentOptions(
+        name="Default Agent",
+        streaming=False,
+        description="This is the default agent that handles general queries and tasks.",
+      ))
     )
     ```
   </TabItem>

python/README.md

@@ -72,7 +72,7 @@ Check out our [documentation](https://awslabs.github.io/multi-agent-orchestrator
 
 
 
-### Installation
+### Core Installation
 
 ```bash
 # Optional: Set up a virtual environment
@@ -81,7 +81,7 @@ source venv/bin/activate  # On Windows use `venv\Scripts\activate`
 pip install multi-agent-orchestrator
 ```
 
-### Usage
+#### Default Usage
 
 Here's an equivalent Python example demonstrating the use of the Multi-Agent Orchestrator with a Bedrock LLM Agent and a Lex Bot Agent:
 
@@ -176,6 +176,19 @@ This example showcases:
 4. Handling of both streaming and non-streaming responses from different types of agents.
 
 
+### Working with Anthropic or OpenAI
+If you want to use Anthropic or OpenAI for classifier and/or agents, make sure to install the multi-agent-orchestrator with the relevant extra feature.
+```bash
+pip install multi-agent-orchestrator[anthropic]
+pip install multi-agent-orchestrator[openai]
+```
+
+### Full package installation
+For a complete installation (including Anthropic and OpenAi):
+```bash
+pip install multi-agent-orchestrator[all]
+```
+
 ## 🤝 Contributing
 
 We welcome contributions! Please see our [Contributing Guide](https://raw.githubusercontent.com/awslabs/multi-agent-orchestrator/main/CONTRIBUTING.md) for more details.

python/setup.cfg

@@ -21,6 +21,13 @@ packages = find:
 python_requires = >=3.11
 install_requires =
     boto3==1.34.151
+
+[options.extras_require]
+anthropic =
+    anthropic==0.32.0
+openai =
+    openai==1.55.0
+all =
     anthropic==0.32.0
     openai==1.55.0
 

python/src/multi_agent_orchestrator/agents/__init__.py

@@ -9,7 +9,13 @@
 from .comprehend_filter_agent import ComprehendFilterAgent, ComprehendFilterAgentOptions
 from .chain_agent import ChainAgent, ChainAgentOptions
 from .bedrock_translator_agent import BedrockTranslatorAgent, BedrockTranslatorAgentOptions
-from .anthropic_agent import AnthropicAgent, AnthropicAgentOptions
+
+try:
+    from .anthropic_agent import AnthropicAgent, AnthropicAgentOptions
+    _ANTHROPIC_AVAILABLE = True
+except ImportError:
+    _ANTHROPIC_AVAILABLE = False
+
 
 __all__ = [
     'Agent',
@@ -24,13 +30,17 @@
     'LexBotAgent',
     'LexBotAgentOptions',
     'AmazonBedrockAgent',
-    'AmazonBedrockAgentOptions'
+    'AmazonBedrockAgentOptions',
     'ComprehendFilterAgent',
     'ComprehendFilterAgentOptions',
     'BedrockTranslatorAgent',
     'BedrockTranslatorAgentOptions',
     'ChainAgent',
     'ChainAgentOptions',
-    'AnthropicAgent',
-    'AnthropicAgentOptions'
 ]
+
+if _ANTHROPIC_AVAILABLE:
+    __all__.extend([
+        'AnthropicAgent',
+        'AnthropicAgentOptions'
+    ])

python/src/multi_agent_orchestrator/classifiers/__init__.py

@@ -1,20 +1,38 @@
 """
 Code for Classifier.
 """
-
-
 from .classifier import Classifier, ClassifierResult
-from .anthropic_classifier import AnthropicClassifier, AnthropicClassifierOptions
 from .bedrock_classifier import BedrockClassifier, BedrockClassifierOptions
-from .openai_classifier import OpenAIClassifier, OpenAIClassifierOptions
+
+try:
+    from .anthropic_classifier import AnthropicClassifier, AnthropicClassifierOptions
+    _ANTHROPIC_AVAILABLE = True
+except Exception as e:
+    _ANTHROPIC_AVAILABLE = False
+
+try:
+    from .openai_classifier import OpenAIClassifier, OpenAIClassifierOptions
+    _OPENAI_AVAILABLE = True
+except Exception as e:
+    _OPENAI_AVAILABLE = False
+
+
 
 __all__ = [
-    "AnthropicClassifier",
-    "AnthropicClassifierOptions",
-    "BedrockClassifier",
-    "BedrockClassifierOptions",
     "Classifier",
     "ClassifierResult",
-    "OpenAIClassifier",
-    "OpenAIClassifierOptions",
+    "BedrockClassifier",
+    "BedrockClassifierOptions"
 ]
+
+if _ANTHROPIC_AVAILABLE:
+    __all__.extend([
+        "AnthropicClassifier",
+        "AnthropicClassifierOptions"
+    ])
+
+if _OPENAI_AVAILABLE:
+    __all__.extend([
+        "OpenAIClassifier",
+        "OpenAIClassifierOptions"
+    ])

python/src/multi_agent_orchestrator/classifiers/classifier.py

@@ -3,7 +3,7 @@
 from typing import Dict, List, Optional
 from dataclasses import dataclass
 from multi_agent_orchestrator.types import ConversationMessage, AgentTypes, TemplateVariables
-from multi_agent_orchestrator.agents import Agent, BedrockLLMAgent, BedrockLLMAgentOptions
+from multi_agent_orchestrator.agents import Agent
 
 
 @dataclass
@@ -13,14 +13,6 @@ class ClassifierResult:
 
 class Classifier(ABC):
     def __init__(self):
-        self.default_agent = BedrockLLMAgent(
-            options=BedrockLLMAgentOptions(
-            name=AgentTypes.DEFAULT.value,
-            streaming=True,
-            description="A knowledgeable generalist capable of addressing a wide range of topics.\
-            This agent should be selected if no other specialized agent is a better fit."
-        ))
-
         self.agent_descriptions = ""
         self.history = ""
         self.custom_variables: TemplateVariables = {}