@promptbook/vercel/esm/typings/src/llm-providers/agent/Agent.d.ts

Promptbook: Turn your company's scattered knowledge into AI ready books

import { BehaviorSubject } from 'rxjs';
import type { AgentBasicInformation, BookParameter } from '../../book-2.0/agent-source/AgentBasicInformation';
import type { string_book } from '../../book-2.0/agent-source/string_book';
import type { LlmExecutionTools } from '../../execution/LlmExecutionTools';
import type { ChatPromptResult } from '../../execution/PromptResult';
import type { Prompt } from '../../types/Prompt';
import type { string_agent_hash, string_agent_name, string_agent_url, string_url_image } from '../../types/typeAliases';
import { AgentLlmExecutionTools } from './AgentLlmExecutionTools';
import type { AgentOptions } from './AgentOptions';
/**
 * Represents one AI Agent
 *
 * Note: [🦖] There are several different things in Promptbook:
 * - `Agent` - which represents an AI Agent with its source, memories, actions, etc. Agent is a higher-level abstraction which is internally using:
 * - `LlmExecutionTools` - which wraps one or more LLM models and provides an interface to execute them
 * - `AgentLlmExecutionTools` - which is a specific implementation of `LlmExecutionTools` that wraps another LlmExecutionTools and applies agent-specific system prompts and requirements
 * - `OpenAiAssistantExecutionTools` - which is a specific implementation of `LlmExecutionTools` for OpenAI models with assistant capabilities, recommended for usage in `Agent` or `AgentLlmExecutionTools`
 * - `RemoteAgent` - which is an `Agent` that connects to a Promptbook Agents Server
 *
 * @public exported from `@promptbook/core`
 */
export declare class Agent extends AgentLlmExecutionTools implements LlmExecutionTools, AgentBasicInformation {
    private _agentName;
    /**
     * Name of the agent
     */
    get agentName(): string_agent_name;
    /**
     * Description of the agent
     */
    personaDescription: string | null;
    /**
     * The initial message shown to the user when the chat starts
     */
    initialMessage: string | null;
    /**
     * Links found in the agent source
     */
    links: Array<string_agent_url>;
    /**
     * Computed hash of the agent source for integrity verification
     */
    get agentHash(): string_agent_hash;
    /**
     * Metadata like image or color
     */
    meta: {
        fullname?: string;
        image?: string_url_image;
        link?: string;
        title?: string;
        description?: string;
        [key: string]: string | undefined;
    };
    /**
     * Not used in Agent, always returns empty array
     */
    get parameters(): BookParameter[];
    readonly agentSource: BehaviorSubject<string_book>;
    constructor(options: AgentOptions);
    /**
     * Calls the chat model with agent-specific system prompt and requirements with streaming
     *
     * Note: This method also implements the learning mechanism
     */
    callChatModelStream(prompt: Prompt, onProgress: (chunk: ChatPromptResult) => void): Promise<ChatPromptResult>;
}
/**
 * TODO: [🧠][😰]Agent is not working with the parameters, should it be?
 */