Skip to main content

Spec Workflow Description

Introduction​

CloudBase AI ToolKit features a built-in Kiro-style Spec workflow, designed to make the AI programming process more engineering-oriented and standardized, enhancing the end-to-end experience of requirements grooming, design, development, and acceptance. It now supports flexible slash command control, allowing users to select the appropriate development mode based on task complexity.

Workflow Philosophy​

  • Engineering: Clarify requirements, design, and tasks to avoid "slot machine-style" vibe coding.
  • Traceable: Every step is documented to facilitate backtracking and team collaboration.
  • AI Collaboration: AI is responsible for grooming requirements, designing solutions, and breaking down tasks, while humans focus on decision-making and review.
  • Automatic Generation: Automatically generates standardized documents such as requirements.md, design.md, tasks.md.
  • Flexible Control: Supports slash command control with intelligent assessment of task complexity.

Workflow Control​

Slash Command System​

CloudBase AI ToolKit provides flexible slash commands to control the workflow:

  • /spec - Forces the use of the full spec process (requirements documentation, technical solution, task breakdown)
  • /no_spec - Skips the spec process and directly executes tasks
  • /help - Display workflow command help
  • Default - AI intelligently determines based on task complexity.

Intelligent Judgment Mechanism​

AI will automatically select the appropriate development mode based on task complexity:

Scenarios for using the spec process:

  • New feature development
  • Complex architecture design
  • Multi-module integration
  • Involves database design
  • Involves user interface design
  • Involves third-party service integration

Scenarios for skipping the spec process:

  • Simple bug fixes
  • Document updates
  • Configuration modification
  • Code refactoring
  • Performance optimization
  • Test case writing

Workflow Process​

Spec Process (/spec or intelligent judgment-based usage)​

  1. Requirements Grooming
    • Adopt the EARS (Easy Approach to Requirements Syntax) method, where AI assists users in grooming requirements and generates the requirements.md file.
  2. Technical Solution Design
    • Based on requirements, AI generates technical architecture, technology selection, interface/database design, etc., forming the design.md file.
  3. Task Breakdown
    • Based on requirements and design, AI breaks down specific development tasks and generates the tasks.md file.
  4. Automated Execution and Tracking
    • AI automatically advances development based on the task list and updates task status in real-time.

No-Spec Process (/no_spec or intelligent judgment-based skipping)​

  1. Direct Task Requirements Analysis
  2. Rapid Execution Plan Development
  3. Direct Task Execution
  4. Maintain Code Quality and Project Standards
  5. Use interactiveDialog for confirmation when necessary

Typical File Structure​

specs/
  your-feature/
    requirements.md  # Requirements Documentation
    design.md        # Technical Solution Design
    tasks.md         # Task Breakdown and Progress

requirements.md Example​

# Requirements Documentation

## Introduction

Briefly describe the requirements background and objectives.

## Requirements

### Requirement 1 - Requirement Name

**User Story:** User story content

#### Acceptance Criteria

1. While <optional precondition>, when <optional trigger>, the <system name> shall <system response>
2. ...

design.md Example​

# Technical Solution Design

- Architecture Diagram (can be created using mermaid)
- Technology Selection Description
- Database/Interface Design
- Security and Testing Strategy

tasks.md Example​

# Implementation Plan

- [ ] 1. Task Information
  - Specific Tasks
  - _Requirement: Related requirement point ID

Usage Example​

Force the Use of Spec Process​

User Input:

/spec Develop a user management system

AI Response:

🎯 Workflow Mode: Spec Process

📋 Criteria: User explicitly specifies to use the /spec command
🔄 Subsequent Process: Will be executed according to standard software engineering practices
  1. Requirements Clarification and Confirmation
  2. Requirements Documentation Design (requirements.md)
  3. Technical Solution Design (design.md)
  4. Task Breakdown (tasks.md)
  5. Task Execution and Status Update

✅ Confirmation: I will begin the requirements clarification phase. Please confirm whether to proceed?

Skip Spec Process​

User Input:

/no_spec Fix the styling issue of the login button

AI Response:

🎯 Workflow Mode: No-Spec Process

📋 Criteria: User explicitly specifies to use the /no_spec command
🔄 Subsequent Process: Will directly analyze and execute tasks
  1. Direct Task Requirements Analysis
  2. Rapid Execution Plan Development
  3. Direct Task Execution
  4. Maintain Code Quality and Project Standards

⚠️ Note: Will skip the requirement documentation and technical solution design phases, directly executing the repair task
✅ Confirmation: I will begin analyzing the login button styling issue. Please confirm whether to proceed?

Intelligent Judgment (Default)​

User Input:

Develop a complex e-commerce shopping cart feature

AI Response:

🎯 Workflow Mode: Spec Process

📋 Criteria: The task involves complex feature development (e-commerce shopping cart)
🔄 Subsequent Process: Will be executed according to standard software engineering practices
  1. Requirements Clarification and Confirmation
  2. Requirements Documentation Design (requirements.md)
  3. Technical Solution Design (design.md)
  4. Task Breakdown (tasks.md)
  5. Task Execution and Status Update

✅ Confirmation: I will begin the requirements clarification phase. Please confirm whether to proceed?

Integration with AI IDE​

  • Supports mainstream AI IDEs such as Cursor, Claude Code, WindSurf, CodeBuddy, etc.
  • Through built-in rules and MCP tools, AI can automatically identify and execute Spec workflows.
  • Users only need to describe requirements in natural language, and the AI automatically generates and maintains Spec documentation.
  • Supports slash command control, offering flexible development mode options

Advantages​

  • Standardized development process: Reduces communication and collaboration costs
  • Flexible control: Selects the appropriate development mode based on task complexity
  • Intelligent judgment: AI automatically determines whether the spec process needs to be used
  • Facilitates team collaboration: Requirement traceability and document management
  • Support for automated testing: Integrates with continuous integration processes
  • Making AI truly the "assistant engineer" in the development workflow

Best Practices​

Selecting the appropriate command​

Scenarios for using /spec:

  • New feature development
  • Complex architecture design
  • Multi-module integration
  • Involves database design
  • Involves user interface design
  • Involves third-party service integration

Scenarios for using /no_spec:

  • Simple bug fixes
  • Document updates
  • Configuration modification
  • Code refactoring
  • Performance optimization
  • Test case writing

Intelligent Judgment Usage​

Recommendation:

  • When uncertain, you may omit specifying commands and let AI intelligently determine.
  • Observe the basis of AI's judgment to understand its logic
  • If the judgment is inaccurate, you can manually specify the command.

Quality Assurance​

Notes:

  • /no_spec mode still needs to follow project specifications.
  • For complex tasks, it is recommended to use /spec to ensure quality.
  • Important operations require user confirmation in any mode.
  • Maintain code quality and maintainability

Typical Usage​

  1. User enters requirements in the AI IDE (can choose to use slash commands)
  2. AI selects the appropriate development mode based on commands or intelligent judgment.
  3. If using the spec process, AI automatically generates requirements.md, design.md, tasks.md under specs/your-feature/
  4. After user confirmation, AI advances development according to the process and continuously updates task status.
  5. After development is complete, all documents are available for review, archiving, and subsequent maintenance.

For further information on the Spec workflow or if you encounter any issues, please refer to the official documentation or join the technical exchange group.