MockProvider
Class: MockProvider
Defined in: src/adapters/llm/MockProvider.ts:103
Implements
Constructors
Constructor
new MockProvider(
options?):MockProvider
Defined in: src/adapters/llm/MockProvider.ts:114
Parameters
options?
MockProviderOptions = {}
Returns
MockProvider
Properties
name
readonlyname:string
Defined in: src/adapters/llm/MockProvider.ts:104
Implementation of
Methods
complete()
complete(
req):Promise<LLMResponse>
Defined in: src/adapters/llm/MockProvider.ts:156
Parameters
req
Returns
Promise<LLMResponse>
Implementation of
realistic()
staticrealistic(options?):MockProvider
Defined in: src/adapters/llm/MockProvider.ts:148
Convenience factory for the playground / Lens demo defaults: a real-feel mock with 3–8 s of "thinking" before the response and 30–80 ms per streamed word. Lets users observe pause/resume, streaming, and tool dispatch happening live without hitting a paid API.
Parameters
options?
MockProviderOptions = {}
Returns
MockProvider
resetReplies()
resetReplies():
void
Defined in: src/adapters/llm/MockProvider.ts:137
Reset the scripted-replies cursor. Useful when reusing one
MockProvider instance across multiple test scenarios — each
scenario can provider.resetReplies() to start from replies[0]
again. No-op when replies was not supplied.
Returns
void
stream()
stream(
req):AsyncIterable<LLMChunk>
Defined in: src/adapters/llm/MockProvider.ts:167
Streaming mode — emits the response content word-by-word so consumers (Lens commentary, chat UIs) can render tokens as they arrive. Tool calls land all at once on the final chunk because that is how real providers (OpenAI, Anthropic) deliver them too.
Parameters
req
Returns
AsyncIterable<LLMChunk>