Variable: Baleybots

const Baleybots: object

Defined in: packages/core/src/index.ts:661

Type Declaration

conditional

conditional: (config) => Processable<unknown, boolean>

A yes/no decision, answered by whoever is configured to answer it.

A condition is z.boolean().describe(goal) — a constrained schema — so a bounded-decision provider can answer it natively and a model can answer it as structured output. Both are answers to the same question; neither is the primary path.

Which one runs is resolved when the conditional runs, not when it is built, so a pipeline defined before setDefaults() still finds the model, and one invocation can override it through ProcessOptions.defaults.

Parameters

config

ConditionalConfig

Returns

Processable<unknown, boolean>

Example

const isSpam = conditional({ name: 'spam-filter', goal: 'Is this email spam?' });
await isSpam(email);   // boolean, from the decision provider or a model

create

create: <TOutputSchema, TTools>(config) => Baleybot<TOutputSchema, TTools>

Create a new Baleybot (static factory method for convenience)

When output is undefined or omitted, the output type is inferred as string.

Type Parameters

TOutputSchema

TOutputSchema = undefined

TTools

TTools extends Record<string, Processable<unknown, unknown> | ToolDefinition<(...args) => unknown> | ZodToolDefinition<any, any>> = Record<string, never>

Parameters

config

BaleybotConfig<TOutputSchema, TTools>

Returns

Baleybot<TOutputSchema, TTools>

Example

// Without model - auto-selects based on available API keys
const bot = Baleybot.create({
  name: 'my-bot',
  goal: 'Help users'
  // No model specified - auto-selects!
});
const result: string = await bot.process('Hello');

// With explicit model
const bot2 = Baleybot.create({
  name: 'my-bot',
  goal: 'Help users',
  model: 'gpt-5.6-luna'
});
const result2: string = await bot2.process('Hello');

// With output - returns typed output
const bot3 = Baleybot.create({
  name: 'my-bot',
  goal: 'Help users',
  model: 'gpt-5.6-luna',
  output: Output.object({ schema: z.object({ sentiment: z.string() }) })
});
const result3: { sentiment: string } = await bot3.process('Hello');