sharptoolbox/ontology-driven-dev

本体驱动业务系统开发技能：需求探索→本体建模→应用构建三步法，基于七模型本体 YAML 与 code-paas 技术底座。支持 WorkBuddy / Claude Code / Codex。

What it solves

This project provides a structured methodology and technical foundation for transforming business requirements into fully functional management systems. It solves the common problem of "model-code misalignment," where the database schema, APIs, and business logic drift apart from the original requirements and the conceptual model.

How it works

The system employs a three-stage pipeline driven by an ontology-based approach:

  1. Requirement Exploration: A rigorous 8-stage process that uses specific prompts to generate a Software Requirement Specification (SRS). It includes mandatory "human-in-the-loop" checkpoints to ensure the AI does not proceed without user confirmation.
  2. Ontology Modeling: The SRS is converted into seven specific YAML models (M1-M7 and MU) covering objects, behaviors, rules, subjects, processes, reports, and UI. This ensures the semantic source of truth is centralized in the YAML files.
  3. Application Construction: Using a provided technical base (code-paas), the system generates a browser-based (BS) application. The technical base is available in both Python (Flask/SQLite) and Go (Gin/PostgreSQL) versions, integrating a React/TS frontend.

Who it’s for

Developers and product managers using AI agents (such as Claude Code, Cursor, or Aider) who want to build business management systems with strict traceability from requirements to code.

Highlights

  • Strict Traceability: Every feature is traceable back to a specific item in the requirement document.
  • Model-Driven Development: Database tables, interfaces, and permissions are generated directly from the ontology YAMLs, preventing logic duplication.
  • Mandatory Human Gates: Forces a pause for human confirmation at each stage of requirement exploration.
  • AI-Native Integration: Generated applications include a built-in AI chat sidebar that utilizes an ontology registry for tool calling and secure SQL execution.
  • Multi-Agent Compatibility: Designed to work across various AI coding tools including Claude Code, Codex, Cursor, and Cline.

Related

  • Dispatch
    Integrating UML Modeling into AI-Driven Development Life Cycles

    An exploration of AI-DLC-UML, a framework designed to bridge the AI-driven software development workflow with structured UML modeling for collaborative design.

  • Project
    dbt-labs/dbt-charts

    A declarative YAML-based framework that wraps SQL to create auditable, version-controlled dashboards and reports, specifically optimized for both human auditors and AI agents.

  • Project
    steedos/steedos-platform

    An AI-native infrastructure that uses a universal metadata standard to enable the instant generation of enterprise software from text descriptions.

  • Project
    microsoft/conductor

    Conductor is a Microsoft‑maintained CLI that lets you define deterministic, version‑controlled multi‑agent AI workflows in YAML and run them with providers like GitHub Copilot or Anthropic Claude. It offers parallel execution, sub‑workflow composition, human‑in‑the‑loop gates, a web dashboard, a TUI fleet manager, and OpenTelemetry tracing, all installable via a one‑liner script or standard Python tools.

  • Project
    IBM/prompt-declaration-language

    A declarative YAML-based language for creating reliable, composable LLM prompts and integrating them into software systems with support for control flow and code execution.