Version 5.2.0

New Features and Improvements

No-Code Editor & WebApp Generator

  • Added the ability to link agents directly from the no-code editor, enabling users to reference agents visually during modeling.

  • Extended the WebApp generator to automatically generate these agent links inside the generated application.

Agent Framework

  • Introduced a flexible Action abstraction for agent state machines, including CustomCodeAction, AgentReply, and LLMReply.

  • Refactored state Body to store structured action sequences and generate cleaner, more maintainable code.

  • Added AgentComponent to the GUI metamodel for visual agent representation.

  • Extended the React dashboard generator to render agent components in the generated frontend.

  • Enhanced agent code-generation templates with improved platform/LLM initialization and support for speech input/output modalities.

REST API Generator

  • Improved foreign key validation in the generated create functions, ensuring related entities are checked before main entity creation.

  • Enhanced handling of associations and relationships, including multiplicity, bidirectional references, and consistent CRUD operations.

  • Improved formatting and readability of generated CRUD functions for better maintainability.

  • Added consistent count endpoints and standardized naming conventions across all generated routes.

  • Enhanced API responses with detailed query metadata, including pagination and filtering information.

  • Added more robust error handling, including structured responses for invalid HTTP methods and runtime issues.

  • Updated the FastAPI template with advanced database connection pooling and improved initialization patterns.

Pydantic Models

  • Improved class generation by sorting models by inheritance, ensuring parent classes are created before children.

  • Removed duplicated inherited attributes and improved overall naming and import consistency.

Validation System

  • Added the centralized validate() method to DomainModel and updated backend endpoints to use structured validation instead of exception catching.

  • Removed redundant validation calls — the backend is now the single source of truth for validation.

AI Assistant & Copilot Instructions

This pull request introduces comprehensive AI assistant and Copilot instruction guides for the BESSER backend repository, ensuring that both human and AI contributors follow consistent, high-quality practices. The instructions cover repository structure, coding standards, documentation expectations, and contribution workflows, and are provided in two synchronized files for compatibility with both GitHub Copilot and Cursor IDE.

Documentation and contribution guidelines:

  • Added .github/copilot-instructions.md with detailed instructions for GitHub Copilot, outlining repository structure, coding standards, documentation requirements, and step-by-step guides for creating new generators and sub-DSLs.

  • Added .cursorrules as a synchronized copy of the Copilot instructions, tailored for Cursor IDE compatibility, ensuring that AI assistant suggestions are aligned across different tools.

Fixes