@promptbook/remote-client/umd/src/book-2.0/agent-source/AgentModelRequirements.d.ts

Promptbook: Create persistent AI agents that turn your company's scattered knowledge into action

import type { LlmToolDefinition } from '../../types/LlmToolDefinition';
import type { string_agent_url } from '../../types/string_agent_url';
import type { string_knowledge_source_link } from '../../types/string_knowledge_source_content';
import type { chococake } from '../../utils/organization/really_any';
/**
 * Model requirements for an agent
 *
 * This is like "compiled" version of agent source
 */
export type AgentModelRequirements = {
    /**
     * The system message that defines the agent's behavior and personality
     */
    readonly systemMessage: string;
    /**
     * Text appended to every user prompt after the user's input and before tools/messages.
     *
     * This allows commitments to emphasize important rules or reminders that should be
     * visible in the prompt in addition to the system message.
     */
    readonly promptSuffix: string;
    /**
     * The model name to use for this agent
     */
    readonly modelName: string;
    /**
     * List of imported agent URLs
     */
    readonly importedAgentUrls?: ReadonlyArray<string_agent_url>;
    /**
     * List of imported file URLs or paths
     */
    readonly importedFileUrls?: ReadonlyArray<string>;
    /**
     * Optional list of knowledge source links that the agent can use
     */
    readonly knowledgeSources?: ReadonlyArray<string_knowledge_source_link>;
    /**
     * List of sample conversations (question/answer pairs)
     */
    readonly samples?: ReadonlyArray<{
        question: string | null;
        answer: string;
    }>;
    /**
     * Temperature for the agent's responses, controlling randomness
     */
    readonly temperature?: number;
    /**
     * Top-p sampling value for the agent's responses
     */
    readonly topP?: number;
    /**
     * Top-k sampling value for the agent's responses
     */
    readonly topK?: number;
    /**
     * Tools available for the agent
     */
    readonly tools?: ReadonlyArray<LlmToolDefinition>;
    /**
     * Optional list of MCP servers that the agent can connect to
     */
    readonly mcpServers?: ReadonlyArray<string>;
    /**
     * The URL of the parent agent to inherit from, or null if there is no parent (i.e., do not inherit from any agent)
     */
    readonly parentAgentUrl: string_agent_url | null;
    /**
     * Is the agent closed to modification by conversation (i.e., it will not learn from interactions and its source code will remain static during conversation)
     */
    readonly isClosed: boolean;
    /**
     * Arbitrary metadata storage for commitments
     * Each commitment can store its own data here
     */
    readonly _metadata?: Record<string, chococake>;
};