MemNet AI Python SDK

The MemNet AI Python SDK provides a Python interface for interacting with the MemNet AI API, including basic request clients and enhanced features.

Features

• Basic API request client
• Memory management functionality
• Thinking functionality
• Dream functionality
• Common sense functionality
• Enhanced client (with LLM integration)

Get API Key

Obtain API Key from MemNetAI dashboard

Installation

pip install memnetai-python-sdk

Quick Start

Basic Client Usage

from memnetai import MemNetAIClient
from memnetai import Message

# Initialize client
client = MemNetAIClient(
    base_url="https://api.memnetai.com",
    api_key="your_api_key"
)

messages = [
    Message(role="user", content="What is artificial intelligence?", character="default")
]

# 1. Memory
response = client.memories(
    memory_agent_name="default",
    namespace="default",
    messages=messages
)
print("Memory result:", response)

# 2. Recall
response = client.recall(
    memory_agent_name="default",
    namespace="default",
    query="What is artificial intelligence?",
)
print("Recall result:", response)

# 3. Thinking functionality
response = client.think(
    memory_agent_name="default",
    namespace="default",
)
print("Thinking result:", response)

# 4. Dream functionality
response = client.dream(
    memory_agent_name="default",
    namespace="default",
)
print("Dream result:", response)

# 5. Common sense functionality
response = client.common_sense(
    common_sense_database_id="default",
    common_sense_text="Science",
)
print("Common sense result:", response)

Enhanced Client Usage

from memnetai import MemNetAIClientPlus
from memnetai import OpenAIConfig


def test_memnetaiclientplus():
    # Configure OpenAI parameters
    openai_config = OpenAIConfig(
        memnetai_api_key="your_memnet_api_key",
        memory_agent_name="default",
        base_url="https://api.openai.com/v1",
        api_key="your_openai_api_key",
        model_name="gpt-3.5-turbo",
    )

    # Initialize enhanced client
    client_plus = MemNetAIClientPlus(config=openai_config)

    # Intelligent conversation (automatic memory and recall)
    try:
        while True:
            if not client_plus.input():  # User entered exit command
                break
            response = client_plus.chat()
            print(f"AI response: {response}")
    except KeyboardInterrupt:
        print("\nClosing program...")
    finally:
        client_plus.close()


if __name__ == "__main__":
    test_memnetaiclientplus()

API Documentation

1. Basic Client (MemNetAIClient)

Initialization Parameters

MemNetAIClient(api_key, base_url="https://api.memnetai.com")

Parameter	Type	Required	Default	Description
api_key	str	Required	None	MemNet AI API key
base_url	str	Optional	"https://api.memnetai.com"	MemNet AI API base URL

1.1 Memory Management (memories)

memories(memory_agent_name, namespace, messages, language="zh-CN", is_third_person=0, metadata="", async_mode=0)

Parameter	Type	Required	Default	Description
memory_agent_name	Optional[str]	Required	None	Memory agent name
messages	List[Message]	Required	None	Message list containing memory content to be saved
namespace	str	Optional	default	Namespace for isolating memories from different applications
language	str	Optional	"zh-CN"	Expected language for memory summary
is_third_person	int	Optional	0	Whether to use third person (0: no, 1: yes)
metadata	str	Optional	""	Metadata information
async_mode	int	Optional	0	Whether to enable asynchronous memory (0: synchronous, 1: asynchronous)

1.2 Recall Function (recall)

recall(memory_agent_name, namespace, query, character="User", recall_deep=1, is_include_linked_new_memories_from_invalid=0, is_using_associative_thinking=1, is_using_common_sense_database=1, is_using_global_common_sense_database=1, is_using_memory_agent_common_sense_database=0, common_sense_database_id_list=None, is_returning_detailed_memory_info=0)

Parameter	Type	Required	Default	Description
memory_agent_name	Optional[str]	Required	None	Memory agent name
query	str	Required	None	Query content
namespace	str	Optional	default	Namespace for isolating memories from different applications
character	str	Optional	"User"	Character name
recall_deep	int	Optional	1	Recall depth
is_include_linked_new_memories_from_invalid	int	Optional	0	Whether to include new memories linked from invalid memories
is_using_associative_thinking	int	Optional	1	Whether to use associative thinking
is_using_common_sense_database	int	Optional	1	Whether to use common sense database
is_using_global_common_sense_database	int	Optional	1	Whether to use global common sense database
is_using_memory_agent_common_sense_database	int	Optional	0	Whether to use memory agent common sense database
common_sense_database_id_list	Optional[List[str]]	Optional	None	List of common sense database IDs
is_returning_detailed_memory_info	int	Optional	0	Whether to return detailed memory information

1.3 Thinking Function (think)

think(memory_agent_name, namespace, subject="", async_mode=1, is_auto=0)

Parameter	Type	Required	Default	Description
memory_agent_name	Optional[str]	Required	None	Memory agent name
namespace	str	Optional	default	Namespace for isolating memories from different applications
subject	str	Optional	""	Thinking subject, randomly selected from historical memories if not specified
async_mode	int	Optional	1	Asynchronous mode (0: synchronous, 1: asynchronous)
is_auto	int	Optional	0	Whether to auto-think (0: manual, 1: auto)

1.4 Dream Function (dream)

dream(memory_agent_name, namespace, subject="", async_mode=1, is_auto=0)

Parameter	Type	Required	Default	Description
memory_agent_name	Optional[str]	Required	None	Memory agent name
namespace	str	Optional	default	Namespace for isolating memories from different applications
subject	str	Optional	""	Dream subject, randomly selected from historical memories if not specified
async_mode	int	Optional	1	Asynchronous mode (0: synchronous, 1: asynchronous)
is_auto	int	Optional	0	Whether to auto-dream (0: manual, 1: auto)

1.5 Common Sense Function (common_sense)

common_sense(common_sense_database_id, common_sense_text, async_mode=1)

Parameter	Type	Required	Default	Description
common_sense_database_id	str	Required	None	Common sense memory database ID
common_sense_text	str	Required	None	Common sense database text content
async_mode	int	Optional	1	Asynchronous mode (0: synchronous, 1: asynchronous)

1.6 Task Management Methods

All task management methods support the following parameter formats:

Method Name	Parameters	Description
memories_detail	task_id	Get memory task details
memories_progress	task_id	Get memory task progress
memories_all	None	Get all memory tasks
memories_delete	task_id	Delete memory task
think_detail	task_id	Get thinking task details
think_all	None	Get all thinking tasks
think_delete	task_id	Delete thinking task
dream_detail	task_id	Get dream task details
dream_all	None	Get all dream tasks
dream_delete	task_id	Delete dream task
common_sense_detail	task_id	Get common sense task details
common_sense_progress	task_id	Get common sense task progress
common_sense_all	None	Get all common sense tasks
common_sense_delete	task_id	Delete common sense task

2. Enhanced Client (MemNetAIClientPlus)

Initialization Parameters

MemNetAIClientPlus(config)

Parameter	Type	Required	Default	Description
config	Config	Required	None	Configuration object, supports OpenAIConfig, etc.

2.1 Intelligent Conversation (chat)

chat()

Parameter	Type	Required	Default	Description
None	None	None	None	Chat with LLM, automatically handles memory and recall

Return value: str - Complete response content from LLM

2.2 User Input (input)

input(message=None)

Parameter	Type	Required	Default	Description
message	Optional[str]	Optional	None	User input message, obtained from console if not provided

Return value: bool - Whether to continue running (True: continue, False: exit)

2.3 Close Client (close)

close()

Parameter	Type	Required	Default	Description
None	None	None	None	Clear message queue and release resources

3. Configuration Classes

3.1 Basic Configuration (Config)

Config(memnetai_api_key, memory_agent_name, namespace="default", memnetai_base_url="https://api.memnetai.com", window_size=32)

Parameter	Type	Required	Default	Description
memnetai_api_key	str	Required	None	MemNet AI API key
memory_agent_name	str	Required	None	Memory agent name
namespace	str	Optional	"default"	Namespace for isolating memories from different applications
memnetai_base_url	str	Optional	"https://api.memnetai.com"	MemNet AI API base URL
window_size	int	Optional	32	Conversation window size, minimum value is 32

3.2 OpenAI Configuration (OpenAIConfig)

OpenAIConfig(memnetai_api_key, memory_agent_name, base_url, model_name, api_key, namespace="default", memnetai_base_url="https://api.memnetai.com", temperature=0.7, max_tokens=500, window_size=32)

Parameter	Type	Required	Default	Description
memnetai_api_key	str	Required	None	MemNet AI API key
memory_agent_name	str	Required	None	Memory agent name
base_url	str	Required	None	OpenAI API base URL
model_name	str	Required	None	OpenAI model name (e.g., "gpt-3.5-turbo")
api_key	str	Required	None	OpenAI API key
namespace	str	Optional	"default"	Namespace for isolating memories from different applications
memnetai_base_url	str	Optional	"https://api.memnetai.com"	MemNet AI API base URL
temperature	float	Optional	0.7	Model temperature parameter, controls output randomness
max_tokens	int	Optional	500	Maximum number of tokens generated by the model
window_size	int	Optional	32	Conversation window size, minimum value is 32

4. Message Class (Message)

Message(role, content, character="")

Parameter	Type	Required	Default	Description
role	str	Required	None	Role (e.g., "user", "assistant", "system")
content	str	Required	None	Message content
character	str	Optional	""	Character name

Example Applications

Complete Conversation Application

from memnetai import MemNetAIClientPlus
from memnetai import OpenAIConfig

# Configuration parameters
config = OpenAIConfig(
    # MemNet AI configuration
    memnetai_api_key="your_memnet_api_key",
    memory_agent_name="my_agent",
    namespace="my_app",
    
    # OpenAI configuration
    base_url="https://api.openai.com/v1",
    api_key="your_openai_api_key",
    model_name="gpt-3.5-turbo",
    temperature=0.7,
    max_tokens=1000,
    
    # Conversation window configuration
    window_size=64
)

# Create client
client = MemNetAIClientPlus(config)

try:
    print("Welcome to MemNet AI Enhanced Client! Enter 'exit' or 'quit' to exit.")
    while True:
        user_input = input("\nUser: ")
        if user_input.lower() in ['exit', 'quit']:
            print("Thank you for using, goodbye!")
            break
            
        # Add user message and get AI response
        if client.input(user_input):
            response = client.chat()
            print(f"AI: {response}")
            
finally:
    client.close()

Error Handling

MemNetAIClient has built-in the following error handling:

• HTTP errors: Display status code and response content
• Network errors: Prompt when unable to connect to the server
• Timeout errors: Prompt for request timeout
• JSON parsing errors: Prompt when API returns non-JSON data
• Unknown exceptions: Catch all other exceptions and display detailed information

Notes

1. API Key Security: Do not hardcode API keys in your code, it is recommended to use environment variables or configuration files
2. Network Connection: Ensure your network can access MemNet AI and OpenAI API
3. Version Compatibility: This SDK requires Python 3.9+
4. Asynchronous Mode: Most APIs support asynchronous mode, suitable for handling large numbers of requests
5. Window Size: The window_size parameter controls the length of conversation history, it is recommended to adjust according to memory conditions, the minimum value is 32, and memories will be stored once every 16 conversations.