#AI-Generated Information Management System Process

Process task execution is one of XDevelop's core capabilities. Through the "AI-Generated Information Management System" process, requirement descriptions are automatically transformed into a complete, runnable information management system.

The entire process is collaboratively completed by 6 sub-agents, from functional design to frontend and backend code delivery, achieving true AI-driven full-stack development.

#Process Overview

The process consists of the following 6 sub-agents executing sequentially:

1. Functional Description → 2. Data Model & Fields → 3. Function Menu
                                                      ↓
6. Frontend Code ← 5. Backend Code ← 4. Code Definition

• The first three sub-agents handle the design phase — transforming requirements into structured system design documents
• The last three sub-agents handle the development phase — generating specific, runnable business code based on the design documents

#Sub-Agent Working Mechanism

Each sub-agent is bound to a corresponding agent tool (e.g., "Save Functional Description"). The system drives the sub-agents through two sets of prompts:

• System Prompt — Defines the sub-agent's role, responsibilities, and output specifications, telling the AI who you are and what to do
• Generation Message Prompt — Carries specific input data (usually from the previous sub-agent's output), telling the AI to generate content based on this information

Sub-agents reference each other's output data through expressions, for example:

${Project Info.Functional Description}     → Gets the content generated by the functional description node

#Detailed Breakdown of the Six Sub-Agents

#1. Functional Description

• Analyzes requirement descriptions provided by the user, outlining which functional modules the project includes
• Defines the core capabilities of each module
• Clarifies the relationships between modules
• Provides directional guidance for subsequent stages

#2. Data Model & Fields

• Designs the complete database table structure
• Defines field types, primary key and foreign key relationships
• Ensures the data layer can support the implementation of all business functions
• Outputs executable database table creation statements

#3. Function Menu

• Generates the system's menu tree and page list
• Clarifies the entry position and hierarchical relationship of each page
• Provides an interface blueprint for frontend code generation

#4. Code Definition

From a system-wide perspective, it establishes coding standards and constraints for each functional module:

• API naming conventions
• Data validation logic
• Permission control strategies
• Exception handling standards

This sub-agent does not write specific code but rather establishes rules, ensuring that the subsequently generated backend and frontend code follows unified standards.

#5. Backend Code

Under the constraints and standards set by the code definition, it generates based on the data model:

• API routes (RESTful API)
• Database operations (CRUD)
• Business logic processing
• Exception handling mechanisms
• Permission verification middleware

Transforming design blueprints into real, runnable server-side code.

#6. Frontend Code

Combines the following inputs to generate complete frontend page code:

• Function Menu Tree — Interface structure
• Code Definition — API specifications
• Functional Description — Business logic

Generated content includes page layout, component interaction, data binding, style rendering, etc.

#Process Characteristics

#Expression References, Flexible Orchestration

Sub-agents reference the output data of other nodes through expressions (e.g., ${Project Info.Functional Description}), rather than through a fixed chain of delivery. Each sub-agent can reference the output of any one or more preceding nodes as needed.

#Multi-Format Output

Based on the business scenario, sub-agents can output data in different formats:

• JSON — Structured data
• Markdown — Document content
• HTML — Page fragments

Not limited to a single format, flexibly adapting to different scenarios.

#Single Responsibility, Replaceable

Each sub-agent does only one thing and does it well, making it easy to optimize or replace individual stages later.

#Constraints Upfront

The code definition stage establishes standards in advance, ensuring that backend and frontend code are consistent in style and controllable in quality.

Figure: Sub-Agent · Frontend Code