@autobe/agent/lib/context/IAutoBeApplicationProps.d.ts

AI backend server code generator

export interface IAutoBeApplicationProps {
    /** The reason of the function call. */
    reason: string;
    /**
     * # Define prompts to translate user planning requirements into messages for internal agents
     *
     * This prompt defines how to convert a user's planning-oriented requirements
     * into a structured message for an internal agent.
     *
     * All content the user provides must be included in the message. However, if
     * some parts of the user's input are inappropriate or insufficient from a
     * planning standpoint, you are allowed to add **supplementary remarks**—but
     * only under strict rules.
     *
     * # Supplementary Remark Rules
     *
     * 1. **Definition** A supplementary remark is additional information that may
     *    differ from the user's original intent. Because of this, **you must
     *    clearly indicate that it is _not_ part of the user’s thinking**.
     * 2. **When to Supplement**
     *
     * - If the user's input reveals a lack of technical understanding (e.g.,
     *   suggesting "put all data into one table"), and the plan is not an MVP or
     *   PoC, it's encouraged to make reasonable additions for a more scalable or
     *   robust structure.
     * - If there are clear gaps in the user's planning logic, you may supplement
     *   the content to ensure completeness.
     *
     * 3. **When Not to Supplement**
     *
     * - If the user's input is vague or ambiguous, **do not assume or add extra
     *   details**. Instead, it’s better to ask the user follow-up questions to
     *   clarify their intent.
     * - If the user has made no comment on design, **do not impose design-related
     *   decisions** (e.g., colors, fonts, tone). However, you may state
     *   explicitly that no design requirements were provided.
     * - Generic advice like "UX should be good" can be omitted unless it adds
     *   value, as such goals are assumed in all services.
     *
     * # Style Guidelines
     *
     * This prompt is delivered to the sub-agent, and several are created for
     * parallel processing of the sub-agent. Additionally, there should be a guide
     * to style, since sub-agents cannot create different styles of documents due
     * to the disconnection of their conversations with each other.
     *
     * For example, there should be a hyperlink to the previous document, the next
     * document, before or after the document, or there should be no more than N
     * headings. The entire content of the document will have requirements, such
     * as maintaining informal or formal language.
     *
     * The style guide should include conventions for Markdown formatting elements
     * such as headings, lists, and tables. Additionally, it should define
     * expectations regarding document length and overall composition. When
     * describing structural guidelines, include a template to illustrate the
     * recommended format.
     *
     * # Limiting the volume of a document
     *
     * However, do not go beyond the volume guide; each agent only needs to create
     * one page because the agent receiving this document will be created as many
     * as the number of pages.
     */
    userPlanningRequirements?: string;
}