Back to docs

Getting Started

Install H·AI·K·U, create your first intent, and deliver working software

Get H·AI·K·U running in your project and deliver your first feature. H·AI·K·U is the software development profile of the H·AI·K·U Method (Human + AI Knowledge Unification).

Installation

/plugin marketplace add gigsmart/haiku-method
/plugin install haiku --scope project

The Two Commands You Need

CommandWhat it does
/haiku:haiku-startCreate an intent — define what you're building and select a studio
/haiku:haiku-pickupRun the stage pipeline — autonomous execution through each stage

Your First Feature

Step 1: Create an Intent

/haiku:haiku-start

The AI guides you through:

  1. What you want to build
  2. Studio selection — software (git + PRs) or ideation (filesystem + local delivery)
  3. Stage mode — single-stage (all disciplines merged) or multi-stage (sequential progression)
  4. Success criteria — how you'll know it's done
  5. Units — how to break down the work (for complex features)

Example session:

User: /haiku:haiku-start
AI: What do you want to build?
User: Add user authentication with email/password
AI: [Asks clarifying questions via interactive prompts]
AI: Here are the success criteria I captured...
AI: Intent created! Run /haiku:haiku-pickup to start.

Step 2: Run Stages

/haiku:haiku-pickup

The AI now works through the stages defined by your studio. For the software studio, stages progress through:

  1. Inception — Understand the problem, elaborate into units
  2. Development — Plan, build, and review each unit
  3. Security — Threat modeling and vulnerability assessment (if applicable)
  4. Operations — Deployment and monitoring (if applicable)

Each stage has its own hats (roles) defined as files in the stage's hats/ directory. The AI transitions through hats automatically, running quality gates at each step.

You can watch, intervene if needed, or let it run.

Step 3: Continue After Context Resets

If the session runs long, the AI will suggest clearing context:

AI: "Context getting full. Run /clear to continue."
User: /clear
User: /haiku:haiku-pickup

Your progress is preserved — the AI picks up where it left off.

Key Commands

CommandPurpose
/haiku:haiku-startCreate a new intent
/haiku:haiku-pickupRun the stage pipeline
/haiku:haiku-gate-reviewPre-delivery code review — catches issues before external CI/bots
/haiku:haiku-pickup [slug]Resume an existing intent after a break
/haiku:haiku-quickQuick mode for small tasks — skip full pipeline
/haiku:haiku-zapZero-ceremony hat loop — typo/one-liner work, no intent or workflow scaffolding
/haiku:haiku-reset-intentWipe the whole intent (every stage, every branch) and recreate from the preserved title/description
/haiku:haiku-reset-stageWipe a single stage (units, outputs, artifacts, branch) so the agent re-runs it from scratch — other stages stay put
/haiku:haiku-autopilotFull autonomous workflow — create, run, review, deliver

Example: Complete Workflow

User: /haiku:haiku-start
AI: What do you want to build?
User: Add a dark mode toggle to the settings page
AI: [Guides through requirements and criteria]
AI: Intent created!

User: /haiku:haiku-pickup
AI: [Works autonomously through inception → development → review]
AI: Intent complete! All criteria satisfied.

User: Great, let's create a PR
AI: [Creates PR with summary of changes]

After Delivery

When your intent completes, the engine reflects on the run automatically before sealing it — synthesizing what happened along the active studio's reflection dimensions. There's no command to invoke; the reflection artifact is written as part of intent completion.

What's Different from AI-DLC?

If you used AI-DLC before the H·AI·K·U rebrand, see the Migration Guide for a complete map of changes. The key shifts:

  • Studios replace workflows — named lifecycle templates that define which stages run
  • Stages replace standalone hats — each stage has its own hats/ directory with per-hat files
  • Persistence adapters — git (PRs) or filesystem (local), configured per studio
  • .haiku/ replaces .ai-dlc/ as the project directory
  • /haiku:* replaces /ai-dlc:* as the command prefix

Next Steps