Class●Since v1.0

OpenAIModerationMiddleware

Moderate agent traffic using OpenAI's moderation endpoint.

OpenAIModerationMiddleware(
  self,
  *,
  model: ModerationModel = 'omni-moderation-latest',
  check_input: bool = True,
  check_output: bool = True,
  check_tool_results: bool = False,
  exit_behavior: Literal['error', 'end', 'replace'] = 'end',
  violation_message: str | None = None,
  client: OpenAI | None = None,
  async_client: AsyncOpenAI | None = None
)

Bases

AgentMiddleware[AgentState[Any], Any]

Parameters

Name	Type	Description
`model`	`ModerationModel`	Default:`'omni-moderation-latest'` OpenAI moderation model to use.
`check_input`	`bool`	Default:`True` Whether to check user input messages.
`check_output`	`bool`	Default:`True` Whether to check model output messages.
`check_tool_results`	`bool`	Default:`False` Whether to check tool result messages.
`exit_behavior`	`Literal['error', 'end', 'replace']`	Default:`'end'` How to handle violations (`'error'`, `'end'`, or `'replace'`).
`violation_message`	`str \| None`	Default:`None` Custom template for violation messages.
`client`	`OpenAI \| None`	Default:`None` Optional pre-configured OpenAI client to reuse. If not provided, a new client will be created.
`async_client`	`AsyncOpenAI \| None`	Default:`None` Optional pre-configured AsyncOpenAI client to reuse. If not provided, a new async client will be created.

Constructors

constructor

__init__

Name	Type
model	ModerationModel
check_input	bool
check_output	bool
check_tool_results	bool
exit_behavior	Literal['error', 'end', 'replace']
violation_message	str \| None
client	OpenAI \| None
async_client	AsyncOpenAI \| None