Share via

ChatPrompt Class

Core class for interacting with AI models through a prompt-based interface.

Handles message processing, function calling, and plugin execution. Provides a flexible framework for building AI-powered applications.

Initialize ChatPrompt with model and optional functions/plugins.

Constructor

ChatPrompt(model: AIModel, *, functions: list[microsoft.teams.ai.function.Function[Any]] | None = None, plugins: list[microsoft.teams.ai.plugin.AIPluginProtocol] | None = None)

Parameters

Name Description
model
Required

AI model implementation for text generation
functions
Required

Optional list of functions the model can call
plugins
Required

Optional list of plugins for extending functionality

Keyword-Only Parameters

Name Description
functions
Default value: None
plugins
Default value: None

Methods

send

Send a message to the AI model and get a response.
with_function

Add a function to the available functions for this prompt.

Can be called in three ways:

  1. with_function(function=Function(...))
  2. with_function(name=..., description=..., parameter_schema=..., handler=...)
  3. with_function(name=..., description=..., handler=...) - for functions with no parameters
with_plugin

Add a plugin to the chat prompt.

send

Send a message to the AI model and get a response.

async send(input: str | UserMessage | ModelMessage | SystemMessage | FunctionMessage, *, memory: Memory | None = None, on_chunk: Callable[[str], Awaitable[None]] | Callable[[str], None] | None = None, instructions: str | SystemMessage | None = None) -> ChatSendResult

Parameters

Name Description
input
Required

Message to send (string will be converted to UserMessage)
memory
Required

Optional memory for conversation context
on_chunk
Required

Optional callback for streaming response chunks
instructions
Required

Optional system message to guide model behavior

Keyword-Only Parameters

Name Description
memory
Default value: None
on_chunk
Default value: None
instructions
Default value: None

Returns

Type Description

ChatSendResult containing the final model response

with_function

Add a function to the available functions for this prompt.

Can be called in three ways:

  1. with_function(function=Function(...))
  2. with_function(name=..., description=..., parameter_schema=..., handler=...)
  3. with_function(name=..., description=..., handler=...) - for functions with no parameters
with_function(function: Function[T]) -> Self

Parameters

Name Description
function

Function object to add (first overload)

Default value: None
name
Required

Function name (second and third overload)
description
Required

Function description (second and third overload)
parameter_schema
Required

Function parameter schema (second overload, optional)
handler
Required

Function handler (second and third overload)

Keyword-Only Parameters

Name Description
name
Default value: None
description
Default value: None
parameter_schema
Default value: None
handler
Default value: None

Returns

Type Description

Self for method chaining

with_plugin

Add a plugin to the chat prompt.

with_plugin(plugin: AIPluginProtocol) -> Self

Parameters

Name Description
plugin
Required

Plugin to add for extending functionality

Returns

Type Description

Self for method chaining