Class StreamingRunnerV3

A Runner encapsulates one strategy for the LLM call → parse → execute loop.

Implements

common.apis.cortex.Runner

Index

Constructors

Properties

history id verbose

Methods

getOutputModule getPromptFormat run warmup

Constructors

constructor

new StreamingRunnerV3(options): common.apis.cortex.StreamingRunnerV3
Parameters
- options: common.apis.cortex.StreamingRunnerV3Options
Returns common.apis.cortex.StreamingRunnerV3
- Defined in packages/ts_common/src/apis/cortex/runner/streaming_v3.ts:381

Properties

history

history: StreamingV3LLMRecord[] = []

`Readonly`id

id: "streaming_v3" = 'streaming_v3'

verbose

verbose: boolean

Methods

getOutputModule

getOutputModule(): common.apis.cortex.ContextModule
Returns common.apis.cortex.ContextModule
Implementation of Runner.getOutputModule
- Defined in packages/ts_common/src/apis/cortex/runner/streaming_v3.ts:407

getPromptFormat

getPromptFormat(): common.apis.cortex.RunnerPromptFormat
Returns common.apis.cortex.RunnerPromptFormat
Implementation of Runner.getPromptFormat
- Defined in packages/ts_common/src/apis/cortex/runner/streaming_v3.ts:399

run

run(ctx, maxLoops): Promise<string>
Parameters
- ctx: common.apis.cortex.RunnerContext
- maxLoops: number
Returns Promise<string>
Implementation of Runner.run
- Defined in packages/ts_common/src/apis/cortex/runner/streaming_v3.ts:420

warmup

warmup(): Promise<void>
Fire-and-forget warmup call. Hits the same Cloud Function container as the streaming path, keeping it warm for the user's first real interaction. The server short-circuits before billing/LLM.

Returns Promise<void>
Implementation of Runner.warmup
- Defined in packages/ts_common/src/apis/cortex/runner/streaming_v3.ts:391