@axarai/axar/dist/agent/types.d.ts

TypeScript-based agent framework for building agentic applications powered by LLMs

import { z, ZodSchema } from 'zod';
import { SchemaConstructor } from '../schema';
import { CoreTool, StreamTextResult, DeepPartial, Output, CoreMessage, LanguageModelV1 } from 'ai';
/**
 * Union type representing all possible input/output type specifications.
 * Can be a Zod schema, a schema constructor, or a primitive constructor.
 * Used to define the shape and validation rules for agent inputs and outputs.
 */
export type InputOutputType = ZodSchema | SchemaConstructor | StringConstructor | NumberConstructor | BooleanConstructor;
/**
 * Metadata for tool annotation
 */
export type ToolMetadata = Readonly<{
    name: string;
    description: string;
    method: string;
    parameters: z.ZodObject<any>;
    /** Whether the parameter is a primitive type wrapped in { value: T } */
    isPrimitiveParam?: boolean;
}>;
/**
 * Type helper for processed stream output that handles both string and object types.
 * For string types, it returns string directly.
 * For object types, it returns a deep partial version of the type, allowing for partial objects during streaming.
 *
 * @typeParam T - The type to process. Can be string or any object type.
 */
export type StreamOutput<T> = T extends string ? string : DeepPartial<T>;
/**
 * Stream result that provides both processed and raw stream access
 */
export interface StreamResult<TOutput> {
    /**
     * Processed stream that automatically handles TOutput type.
     * For string outputs, provides string chunks.
     * For object outputs, provides partial objects as they stream.
     */
    stream: AsyncIterable<StreamOutput<TOutput>>;
    /** Raw stream access for advanced usage */
    raw: StreamTextResult<Record<string, CoreTool>, TOutput>;
}
/**
 * Type alias for the experimental output configuration returned by Output.object
 */
export type ExperimentalOutput = ReturnType<typeof Output.object>;
/**
 * Configuration for agent output handling
 */
export interface OutputConfig {
    model: LanguageModelV1;
    messages: CoreMessage[];
    tools: Record<string, CoreTool>;
    maxSteps: number;
    /** Maximum number of tokens to generate */
    maxTokens?: number;
    /** Sampling temperature between 0 and 1 */
    temperature?: number;
    /** Nucleus sampling - sample from the smallest set of tokens whose cumulative probability exceeds topP */
    topP?: number;
    /** Only sample from the top K options for each subsequent token */
    topK?: number;
    /** Penalize tokens based on their presence in the prompt and generated text so far */
    presencePenalty?: number;
    /** Penalize tokens based on their frequency in the generated text so far */
    frequencyPenalty?: number;
    /** Maximum number of retries for failed requests */
    maxRetries?: number;
    /** Tool choice mode - 'auto' or 'none' */
    toolChoice?: 'auto' | 'none';
    experimental_telemetry: {
        isEnabled: boolean;
        functionId: string;
    };
    experimental_output?: ExperimentalOutput;
}
/**
 * Configuration options for the language model.
 */
export interface ModelConfig {
    /** Maximum number of tokens to generate */
    maxTokens?: number;
    /** Sampling temperature between 0 and 1. Use either temperature or topP, not both. */
    temperature?: number;
    /** Nucleus sampling - sample from the smallest set of tokens whose cumulative probability exceeds topP. Use either temperature or topP, not both. */
    topP?: number;
    /** Only sample from the top K options for each subsequent token. Recommended for advanced use cases only. */
    topK?: number;
    /** Penalize tokens based on their presence in the prompt and generated text so far. Value between -2.0 and 2.0. */
    presencePenalty?: number;
    /** Penalize tokens based on their frequency in the generated text so far. Value between -2.0 and 2.0. */
    frequencyPenalty?: number;
    /** Maximum number of retries for failed requests */
    maxRetries?: number;
    /** Maximum number of steps in a conversation */
    maxSteps?: number;
    /** Tool choice mode - 'auto' or 'none' */
    toolChoice?: 'auto' | 'none';
}