agentsociety.cityagent.blocks.social_block

Module Contents

Classes

MessagePromptManager	Manages the creation of message prompts by dynamically formatting templates with agent-specific data.
SocialNoneBlock	NoneBlock
FindPersonBlock	Block for selecting an appropriate agent to socialize with based on relationship strength and context.
MessageBlock	Generate and send messages
SocialBlockParams
SocialBlockContext
SocialBlock	Orchestrates social interactions by dispatching to appropriate sub-blocks.

API

class agentsociety.cityagent.blocks.social_block.MessagePromptManager

Manages the creation of message prompts by dynamically formatting templates with agent-specific data.

Initialization

async get_prompt(memory, step: dict[str, Any], environment_info: str, target: int, template: str)

Generates a formatted prompt for message creation.

Args: memory: Agent’s memory to retrieve status data. step: Current workflow step containing intention and context. target: ID of the target agent for communication. template: Raw template string to be formatted.

Returns: Formatted prompt string with placeholders replaced by agent-specific data.

class agentsociety.cityagent.blocks.social_block.SocialNoneBlock(toolbox: agentsociety.agent.AgentToolbox, agent_memory: agentsociety.memory.Memory)

Bases: agentsociety.agent.Block

NoneBlock

Initialization

name

‘SocialNoneBlock’

description

‘Handle all other cases if you are not trying to determine the social target or send a message to som…’

async forward(context)

Executes default behavior when no specific block matches the intention.

Args: step: Current workflow step with ‘intention’ and other metadata. context: Additional execution context (e.g., agent’s plan).

Returns: A result dictionary indicating success/failure, time consumed, and execution details.

class agentsociety.cityagent.blocks.social_block.FindPersonBlock(toolbox: agentsociety.agent.AgentToolbox, agent_memory: agentsociety.memory.Memory)

Bases: agentsociety.agent.Block

Block for selecting an appropriate agent to socialize with based on relationship strength and context.

Initialization

name

‘FindPersonBlock’

description

‘Find a suitable person to socialize with’

async forward(context: agentsociety.agent.DotDict)

Identifies a target agent and interaction mode (online/offline).

Args: context: Additional execution context (may store selected target).

Returns: Result dict with target agent, interaction mode, and execution status.

class agentsociety.cityagent.blocks.social_block.MessageBlock(toolbox: agentsociety.agent.AgentToolbox, agent_memory: agentsociety.memory.Memory)

Bases: agentsociety.agent.Block

Generate and send messages

Initialization

name

‘MessageBlock’

description

‘Send social message to someone (including online and offline, phone call, social post, etc.)’

async forward(context: agentsociety.agent.DotDict)

Generates a message, sends it to the target, and updates chat history.

Args: context: Execution context (may contain pre-selected target).

Returns: Result dict with message content, target, and execution status.

class agentsociety.cityagent.blocks.social_block.SocialBlockParams

Bases: agentsociety.agent.BlockParams

class agentsociety.cityagent.blocks.social_block.SocialBlockContext

Bases: agentsociety.agent.BlockContext

target: Optional[int]

‘Field(…)’

class agentsociety.cityagent.blocks.social_block.SocialBlock(toolbox: agentsociety.agent.AgentToolbox, agent_memory: agentsociety.memory.Memory, block_params: Optional[agentsociety.cityagent.blocks.social_block.SocialBlockParams] = None)

Bases: agentsociety.agent.Block

Orchestrates social interactions by dispatching to appropriate sub-blocks.

Initialization

ParamsType

None

OutputType

None

ContextType

None

NeedAgent

True

name

‘SocialBlock’

description

‘Do social interactions, for example, find a friend, send a message, and other social activities.’

actions

None

async forward(agent_context: agentsociety.agent.DotDict) → agentsociety.cityagent.sharing_params.SocietyAgentBlockOutput

Main entry point for social interactions. Dispatches to sub-blocks based on context.

Args: step: Workflow step containing intention and metadata. context: Additional execution context.

Returns: Result dict from the executed sub-block.