Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

58 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

kiro-powers-aidlc

AI-Driven Development Lifecycle + Power Orchestration — a Kiro Power that combines the official AI-DLC methodology with automated GitHub sync, multi-power coordination, security extensions, traceability enforcement, and zero-config setup.

Quick Start

Mac / Linux (30 seconds)

# From your project root (must have a git remote):
git clone https://github.com/intraedge-services/kiro-powers-aidlc.git /tmp/aidlc
/tmp/aidlc/scripts/setup-aidlc.sh
rm -rf /tmp/aidlc

Windows — PowerShell (with git)

cd C:\path\to\your-project
git clone https://github.com/intraedge-services/kiro-powers-aidlc.git C:\temp\aidlc
powershell -ExecutionPolicy Bypass -File "C:\temp\aidlc\scripts\setup-aidlc.ps1"
Remove-Item C:\temp\aidlc -Recurse -Force

Windows — PowerShell (no git installed)

# Download and extract the power
Invoke-WebRequest -Uri "https://github.com/intraedge-services/kiro-powers-aidlc/archive/refs/heads/main.zip" -OutFile "$env:USERPROFILE\Downloads\aidlc.zip"
Expand-Archive -Path "$env:USERPROFILE\Downloads\aidlc.zip" -DestinationPath "$env:USERPROFILE\Downloads\aidlc" -Force

# Run setup from your project directory
cd C:\path\to\your-project
powershell -ExecutionPolicy Bypass -File "$env:USERPROFILE\Downloads\aidlc\kiro-powers-aidlc-main\scripts\setup-aidlc.ps1"

# Cleanup
Remove-Item "$env:USERPROFILE\Downloads\aidlc.zip", "$env:USERPROFILE\Downloads\aidlc" -Recurse -Force

That's it. No questions asked — it auto-detects everything from your git remote, package files, and gh CLI.

Then open the project in Kiro and say:

"Using AI-DLC, build me ..."

What Gets Installed

After running the setup script, your project gets:

your-project/
├── .kiro/
│   ├── steering/              # Auto-loaded context for Kiro
│   │   ├── project-config.md  # Your project identity + preferences
│   │   ├── core-workflow.md   # Full AIDLC stage rules
│   │   ├── git-signing.md    # Commit signing enforcement
│   │   ├── power-orchestration.md
│   │   └── github-integration.md
│   ├── hooks/                 # Automated triggers
│   │   ├── spec-to-issues.json     # Stories → GitHub issues
│   │   ├── aidlc-board-sync.json   # Close issues on completion
│   │   ├── pr-review-check.json    # PR review validation
│   │   └── pre-code-gen.json       # Activate powers at stage transitions
│   └── aws-aidlc-rule-details/     # Workflow rule files + extensions
├── aidlc-docs/                # AIDLC artifacts (state, audit log)
└── .aidlc-version             # Installed power version (for upgrades)

Upgrade Detection

The setup script is safe to re-run:

# Already installed and current:
$ ./setup-aidlc.sh
✓ AIDLC already up to date (v1.1.0)

# Installed but outdated:
$ ./setup-aidlc.sh
  Installed: v1.0.0
  Available: v1.1.0
  Upgrade from v1.0.0 to v1.1.0? (y/n):

# Skip prompt:
$ ./setup-aidlc.sh --force

Existing steering files are never silently overwritten — backups are created if content differs.

Team Setup

One person runs setup. Everyone else just clones.

# Person 1 (you):
/tmp/aidlc/scripts/setup-aidlc.sh
git add .kiro/ aidlc-docs/ .aidlc-version
git commit -S -m "chore: add AIDLC workflow config"
git push

# Person 2, 3, ... (teammates):
git clone git@github.com:your-org/your-repo.git
# Done. .kiro/ is already there. Open in Kiro and go.

Prerequisites

Tool Required? Purpose
Kiro IDE Yes AI-powered development environment
gh CLI Recommended GitHub issue sync (install)
Git remote Recommended Auto-detection of org/repo/branch
GPG key Recommended Commit signing (setup guide)

Run gh auth login once for GitHub sync to work. If gh isn't available, the workflow still works — it just skips issue creation.

Setup Script Options

# Override specific values:
./setup-aidlc.sh --board 7 --lead "username"

# Force interactive mode:
./setup-aidlc.sh --interactive

# Skip upgrade prompt:
./setup-aidlc.sh --force

# All flags:
./setup-aidlc.sh --org NAME --repo NAME --board NUMBER --lead USER --lang LANGUAGE --force

What Gets Auto-Detected

Value Source
GitHub Org git remote get-url origin
GitHub Repo git remote get-url origin
Default Branch git symbolic-ref
Team Lead gh api user (current authenticated user)
Project Board gh project list --owner ORG (first board found)
Language package.json → JS/TS, pyproject.toml → Python, pom.xml → Java, etc.
Framework cdk.json → AWS CDK, *.tf → Terraform, etc.
GPG Signing git config --get user.signingkey

Windows / WSL Compatibility

Every platform-specific command in the setup script includes inline comments with Windows equivalents:

# Mac/Linux:
brew install gh

# Windows (WSL): sudo apt install gh
# Windows (native): winget install GitHub.cli

The AIDLC Workflow

When you say "Using AI-DLC, build me X", the workflow guides you through:

Inception Phase (What to build)

Stage Always? Purpose
Workspace Detection Yes Brownfield vs greenfield, feature slug confirmation
Reverse Engineering Brownfield only Architecture docs, component inventory
Requirements Analysis Yes Adaptive depth based on complexity
User Stories Conditional Personas + acceptance criteria
Application Design Conditional Components, services, dependencies
Units Generation Conditional Decomposition for complex systems
Workflow Planning Yes Which stages to execute

Construction Phase (How to build it)

Stage Always? Purpose
Functional Design Conditional Business logic, domain entities, rules
NFR Requirements Conditional Performance, security, scalability
NFR Design Conditional Patterns for non-functional requirements
Infrastructure Design Conditional Cloud resources, deployment architecture
Code Generation Yes Plan → implement → verify
Build and Test Yes Verification instructions

Each stage has quality gates — you approve before moving forward.

Feature-Scoped Traceability

Every AIDLC cycle creates a feature-scoped directory. Previous artifacts are never overwritten:

aidlc-docs/
├── setup-aidlc/               # Feature 1
│   ├── inception/
│   └── construction/
├── user-auth/                 # Feature 2
│   ├── inception/
│   └── construction/
├── aidlc-state.md             # Shared (current feature context)
├── audit.md                   # Shared (append-only)
└── requirements-index.md      # Consolidated index of all features

At Workspace Detection, the agent proposes a feature slug and waits for your confirmation before creating directories.

Extensions

Rule sets that enforce constraints during construction stages:

Always Enforced (no opt-in needed)

Extension Rules What It Enforces
Requirement Versioning REQ-VER-01–05 Feature-scoped directories, requirements index, no-overwrite protection, slug confirmation
Git Commit Signing GIT-SIGN-01–04 GPG/SSH signed commits, guided setup if missing

Opt-In (presented during Requirements Analysis)

Extension Rules What It Enforces
Security Baseline SEC-01–08 OWASP input validation, auth, secrets, encryption, logging, dependencies
Resiliency Baseline RES-* AWS Well-Architected reliability patterns
Property-Based Testing TEST-* Hypothesis/fast-check property-based test generation

Extensions follow the ## Rule PREFIX-NN: format with verification checklists at each stage.

GitHub Sync (via gh CLI)

No MCP power needed. The workflow uses gh CLI directly:

Event What Happens
User stories approved Creates GitHub issues with aidlc:story label
Code generation starts Comments on issue: "🔄 Code Generation Started"
Code generation approved Closes issue: "✅ Complete"

Config in project-config.md:

- **Auto-create Issues**: yes
- **Auto-sync Board**: yes

Set Auto-create Issues: no to disable sync entirely.

Hooks

Pre-built event-driven automation:

Hook Trigger What It Does
spec-to-issues.json PostFileCreate (stories.md) Syncs approved stories to GitHub issues
aidlc-board-sync.json PostToolUse Updates board on stage transitions
pr-review-check.json PreToolUse Validates PR review requirements
pre-code-gen.json PreTaskExec Activates registered powers before code gen

Optional Powers

Register additional powers in project-config.md for enhanced validation:

## Installed Powers Registry

| Category | Power Name | Activate During |
|----------|-----------|------------------|
| infrastructure | kiro-powers-aws-cdk-python | Infrastructure design, code gen |
| diagrams | kiro-powers-diagrams | Architecture docs |
| ci-cd | kiro-powers-circleci | Pipeline validation |
| data-engineering | kiro-powers-aws-data-engineering | Glue/EMR/Athena workloads |

Powers add validation and best-practice guidance but never block progress. If a power isn't installed, the workflow warns and continues.

Skills

Reusable skills bundled with the power:

Skill Purpose
setup-aidlc Install/upgrade AI-DLC in any project — supports Kiro, Cursor, Cline, Q Developer, Copilot, Claude Code
aidlc-analyze Pre-workflow analysis mode — explore domain, constraints, risks
aidlc-explore Mid-workflow investigation — compare approaches, clarify decisions

Skills follow the Agent Skills standard. See steering/skills-convention.md for conventions when creating new skills.

Explore & Analyze Modes

Two thinking-partner modes available anytime:

  • Analyze — Deep-dive before starting. Understand domain, constraints, risks.
  • Explore — Mid-workflow investigation. Compare approaches, clarify decisions.

Project Structure

kiro-powers-aidlc/
├── scripts/
│   ├── setup-aidlc.sh           # Zero-prompt bootstrap (upgrade-aware)
│   ├── sync-stories-to-github.sh # Bulk story creation helper
│   └── init-workspace.sh        # Workspace initialization
├── steering/                     # Loaded by Kiro as context
│   ├── core-workflow.md          # Main AIDLC orchestration rules
│   ├── github-integration.md    # GitHub sync via gh CLI
│   ├── power-orchestration.md   # Multi-power coordination
│   ├── skills-convention.md     # Agent Skills standard for creating skills
│   └── project-config-template.md
├── workflows/                    # Rule-detail files
│   ├── common/                   # Shared utilities (10 files)
│   ├── inception/                # Planning stages (7 files)
│   ├── construction/             # Build stages (6 files)
│   ├── operations/               # Placeholder for future deployment workflows
│   └── extensions/               # Blocking rule extensions
│       ├── core/                 # Always-enforced (requirement versioning)
│       ├── security/             # Security baseline + git signing
│       ├── resiliency/           # AWS reliability patterns
│       └── testing/              # Property-based testing
├── hooks/                        # Event-driven automation (4 files)
├── skills/                       # Reusable AI skills
│   └── setup-aidlc/SKILL.md    # Multi-IDE AIDLC installer
├── templates/
│   └── project-config.md        # Template for manual setup
├── examples/
│   └── sample-project-config.md # Example configuration
├── POWER.md                      # Power manifest
├── package.json
└── README.md

Relationship to awslabs/aidlc-workflows

This power is a superset of the official AWS AIDLC rules:

Aspect awslabs/aidlc-workflows This power
Scope Generic rules for any AI agent Full Kiro-native orchestration
Extensions None 5 extensions with formal verification
GitHub Sync None Automated spec-to-issues + board updates
Power Orchestration None Multi-power coordination
Hooks None 4 pre-built event triggers
Upgrade Detection None Version-aware reinstall
Multi-IDE Support Download zip Skill with IDE auto-detection

The skills/setup-aidlc skill can install the vanilla AWS rules for non-Kiro agents (Cursor, Cline, etc.). Our power's workflow engine is independent — no runtime dependency on the AWS release.

Credits

License

MIT

About

Orchestrator power for structured AI-assisted software development. Manages the full AIDLC workflow and automatically activates other powers at the right stages.

Resources

Stars

Watchers

Forks

Releases

Packages

Used by

Contributors

Languages