Skip to main content
A free-form chat-completions LLM (optionally vision-capable) — the JS analogue of the ask primitive. Given a prompt, optional accessibility/structural text, and zero or more images, [AskModel.ask] returns the model’s response as a plain string. The wire format is OpenAI-compatible chat completions, so any provider config that points at such an endpoint works.

​
Constructor

​
Returns

AskModel

​
Properties

​
string
required
The wire-level model identifier sent in the request body.

​
Methods

​
ask()

Ask the model a question, optionally grounded in accessibility text and/or images.
  • prompt: the question or task to answer.
  • text: optional accessibility-tree (or any) text included as structural context. Pass null / omit to skip.
  • images: zero or more images to attach. Each is encoded as a base64 data URL in a model-supported format and sent as an image_url chat content part. Pass null / omit for none.
Returns the trimmed assistant response on success.

​
Parameters

​
string
required
​
null | string
​
null | Image[]
Type: Image

​
Returns

string

​
askJson()

Ask the model for one JSON value that matches schema. Same prompt, text, and images as [AskModel.ask]. schema is a JSON Schema document. A caller that has it as text parses it before calling. The request sends that document unchanged. "answer" is only the name the chat API requires on the schema; it is not a key in the reply. One call carries one schema. The schema is compiled locally before the request. The reply is parsed and checked against that same schema. A reply that is not JSON, or that does not match, throws. A document that is not valid JSON Schema throws and no request is sent. The provider is also asked for strict mode. The local check is what decides whether the value is returned. additionalProperties and required are sent exactly as the caller wrote them. Throws and sends no request when this model does not support it. [AskModel.jsonSchemaAliases] lists the models that do.

​
Parameters

​
string
required
​
undefined | null | string
required
​
undefined | null | Image[]
required
Type: Image
​
any
required

​
Returns

any

​
availableAliases()

Every LLM model alias accepted by byAlias on this machine, deduplicated and sorted alphabetically. Use to discover what aliases the loaded config (bundled defaults plus any user provider files) advertises.

​
Returns

string[]

​
byAlias()

Resolve a model alias against the loaded config (e.g. "openrouter_gpt_4o_mini" / "openrouter_claude_opus" from the bundled openrouter provider, or any alias declared by a user provider). Throws if the alias is unknown.

​
Parameters

​
string
required

​
Returns

AskModel

​
checkAuth()

Check that the API key works. Throws if it doesn’t. Call right after creating the model so a bad key fails fast, before any UI automation has had a chance to steal focus:

​
default()

First LLM model advertised by the first LLM-capable provider in the loaded configuration whose credentials are currently available. Providers with missing credentials are skipped; throws if no provider remains available.

​
Returns

AskModel

​
jsonSchemaAliases()

LLM aliases whose provider entry sets json_schema to true, using the same auth filter and first-provider-wins rule as [AskModel.byAlias]. [AskModel.availableAliases] still lists every LLM, including models that only serve [AskModel.ask].

​
Returns

string[]