Skip to main content

Interface: TemporalRunOptions<TContext>

openaiAgentsWorkflow.TemporalRunOptions

Type parameters

NameType
TContextundefined

Properties

callModelInputFilter

Optional callModelInputFilter: CallModelInputFilter

Edit system instructions or input items just before calling the model


context

Optional context: TContext

Run context passed to agents and tools


conversationId

Optional conversationId: string

OpenAI conversation ID for multi-turn persistence


maxTurns

Optional maxTurns: number

Maximum agent loop turns before aborting


previousResponseId

Optional previousResponseId: string

Previous OpenAI response ID for conversation continuity


runConfig

Optional runConfig: Object

Runner-level config overrides

Type declaration

NameTypeDescription
groupId?stringGrouping ID for linking traces (e.g., chat thread ID)
handoffInputFilter?(input: HandoffInputData) => HandoffInputData-
inputGuardrails?InputGuardrail[]Input guardrails run inline in the Workflow — callbacks must be deterministic
model?stringModel name override (string only — Model objects can't cross the Workflow/Activity boundary)
modelSettings?ModelSettingsGlobal model settings (temperature, topP, etc.). Non-null values override agent-specific settings.
outputGuardrails?OutputGuardrail<AgentOutputType<unknown>, unknown>[]Output guardrails run inline in the Workflow — callbacks must be deterministic
traceId?stringCustom trace ID
traceIncludeSensitiveData?booleanInclude sensitive data (tool I/O, LLM outputs) in trace spans
traceMetadata?Record<string, string>Additional metadata attached to the trace
tracingDisabled?booleanDisable tracing for this run
workflowName?stringLogical name for the run, used in tracing

session

Optional session: Session

Session state for conversation memory


sessionInputCallback

Optional sessionInputCallback: SessionInputCallback

Customize how session history merges with current turn input


stream

Optional stream: boolean

Stream incremental events as the model responds. Requires a streamingTopic (set via the plugin's modelParams or the runner's defaultModelParams) and a hosted WorkflowStream in the Workflow.

Streaming support is experimental and may change without notice.


tracing

Optional tracing: TracingConfig

Per-run tracing config override