Adaptive AI-Driven Development Life Cycle (AI-DLC) for Cursor

This project implements the Adaptive AI-DLC workflow (adapted from Amazon AI-DLC) for the Cursor IDE.

It transforms the AI from a simple assistant into a structured, workflow-driven team member that "adapts to the work, not the other way around."

Key Principles

• Adaptive Workflow: The AI assesses complexity and risk to determine which stages to execute (e.g., skip detailed design for simple changes).
• Audit Logging: All interactions are logged to aidlc-docs/audit.md (append-only, never overwrite).
• State Tracking: Workflow progress is tracked in aidlc-docs/aidlc-state.md.
• Knowledge Base: Uses detailed rule definitions in .cursor/aidlc-rule-details/ to guide the AI through each phase.
• Reusable Skills: Domain-specific skills in .cursor/skills/ provide targeted guidance for common tasks (TDD, testing, demos).

How to Use

Option 1: Trigger Phrase

Start a new task in Cursor Composer (Cmd+I or Cmd+L) with the trigger phrase:

"Using AI-DLC, [your intent]"

Examples:

• "Using AI-DLC, plan a new user profile integration."
• "Using AI-DLC, refactor the login service."

Option 2: Slash Commands

Type / in Cursor chat to access AI-DLC commands directly:

Command	Purpose
/start-aidlc	Start a new AI-DLC workflow with your intent
/resume-aidlc	Continue from where you left off
/check-status	View current workflow state and progress

Inception Commands (in /inception/):

Command	Purpose
/analyze-requirements	Execute Requirements Analysis stage
/create-user-stories	Generate user stories for features
/plan-workflow	Create execution plan for Construction

Construction Commands (in /construction/):

Command	Purpose
/generate-code	Start code generation for a unit
/run-tests-and-fix	Build, test, and fix failures
/review-code	Code review with AI-DLC compliance

Utility Commands (in /utilities/):

Command	Purpose
/view-audit	Show recent audit log entries
/reset-state	Reset workflow state (with confirmation)

The AI will automatically:

1. Log your request in the audit log.
2. Detect your workspace state (greenfield vs. brownfield).
3. Guide you through the phases: Inception, Construction, and Operations.

Phases

🔵 Inception Phase

Goal: Determine WHAT to build and WHY.

• Workspace Detection: Analyzes current state.
• Reverse Engineering: (Brownfield only) Maps existing code.
• Requirements Analysis: Clarifies intent and scope.
• User Stories: (Conditional) For user-facing features.
• Workflow Planning: Decides the strategy for Construction.

🟢 Construction Phase

Goal: Determine HOW to build it and implement.

• Functional Design: For complex logic.
• NFRs (Performance/Security): If needed.
• Infrastructure Design: If cloud resources needed.
• Code Generation: Plan -> Execute -> Test.
• Build and Test: Final verification.

🟡 Operations Phase

• Placeholder for deployment and monitoring workflows.

Directory Structure

<PROJECT-ROOT>/
├── src/                        # Your application code
├── .cursor/
│   ├── rules/
│   │   └── aidlc-core.mdc      # Main Orchestrator Rule (alwaysApply: true)
│   ├── commands/               # Slash commands for AI-DLC workflow
│   │   ├── start-aidlc.md      # Main workflow entry point
│   │   ├── resume-aidlc.md     # Resume from last state
│   │   ├── check-status.md     # View current state
│   │   ├── inception/          # Inception phase commands
│   │   ├── construction/       # Construction phase commands
│   │   └── utilities/          # Utility commands
│   ├── aidlc-rule-details/     # Detailed workflow definitions
│   │   ├── common/             # Cross-phase guidance
│   │   ├── inception/          # Inception phase details
│   │   ├── construction/       # Construction phase details
│   │   └── operations/         # Operations phase details
│   └── skills/
│       ├── ai_dlc_core/        # Core workflow skills (TDD, demo prep)
│       ├── java_dev/           # Java-specific skills
│       └── angular_dev/        # Angular-specific skills
├── aidlc-docs/                 # AI-DLC Documentation (Generated)
│   ├── aidlc-state.md          # Current workflow status & checklist
│   ├── audit.md                # Audit log of all interactions
│   ├── inception/              # Generated plans and specs
│   ├── construction/           # Design docs and build instructions
│   └── operations/             # Deployment and monitoring docs
└── plans/                       # (Optional) Inception phase outputs
    └── README.md               # Usage guide for plans/

Setup Verification

To ensure AI-DLC is active:

1. Open .cursor/rules/aidlc-core.mdc and verify it has alwaysApply: true.
2. Type / in Cursor chat - you should see AI-DLC commands (start-aidlc, resume-aidlc, etc.).
3. Check .cursor/skills/ for available domain-specific skills (Java, Angular, TDD workflows, etc.).

Supported Skills

The project includes domain-specific skills that augment the base workflow:

Skill	Location	Purpose
TDD Workflow	.cursor/skills/ai_dlc_core/tdd-workflow/	Red-Green-Refactor cycle guidance for Construction phase
Demo Preparation	.cursor/skills/ai_dlc_core/demo-preparation/	Prepares features for sprint demos with scripts and verification
Inception Elaboration	.cursor/skills/ai_dlc_core/inception-elaboration/	Guides AI through complex story requirement elaboration
Java Unit Testing	.cursor/skills/java_dev/unit-testing/	Comprehensive unit tests for Java Spring Boot services
Angular Component Testing	.cursor/skills/angular_dev/component-testing/	Angular component testing with Jasmine/Karma

License

MIT-0 (See LICENSE). Based on awslabs/aidlc-workflows.