Info
Unlike the workflow commands (/moai plan, /moai run, /moai sync), the utility commands specialize in fast automation and problem solving.

The Execution Surface of Agentic Loop Engineering

This section is where v3’s second pillar, agentic loop engineering, actually runs. /moai loop is a diagnostic loop that iterates until the issue queue found by diagnostic tools is empty, and /moai goal is a condition-declared loop where you declare a completion condition and the session works on its own until the condition holds. /moai fix is the one-shot (single-pass) preset in the same family. The observations left behind as the loop runs become raw material for harness learning — the entrance to the recursive cycle where loops accumulate observations, the harness learns, and the guidance evolves.

Command Comparison

Command	Purpose	Execution style	When to use
`/moai`	Fully autonomous automation	The whole pipeline from SPEC creation to documentation	When you want to hand over a new feature end to end
`/moai goal`	Condition-declared autonomous loop	Keeps taking turns until the completion condition holds	When you want to declare an end state like “until all tests pass”
`/moai loop`	Iterative fix loop	Repeats diagnose → fix → verify	When you want to knock out many errors at once
`/moai fix`	One-shot auto-fix	Diagnose → fix → done (once)	When you want to fix lint or type errors quickly
`/moai gate`	Pre-commit quality gate	Lint · format · type · test in parallel (<30s)	When you want a fast pass/fail check before committing
`/moai mx`	@MX code annotations	Codebase scan → @MX tag insertion	When you want to plant AI-context code anchors and danger markers
`/moai clean`	Dead-code removal	Static analysis → usage graph → safe removal	When you want to clean up unused code
`/moai codemaps`	Architecture doc generation	Codebase scan → auto-generated structure docs	When you want to produce project architecture docs
`/moai review`	Multi-lens code review	Security · performance · quality · UX verdict + @MX compliance check	When you want a multi-lens review of a PR/changeset
`/moai e2e`	E2E testing	web/mobile/desktop auto-detection + CLI-first runs	When you want to create and run user-journey E2E tests
`/moai feedback`	Submit feedback	Auto-creates a GitHub issue	When sending a bug report or improvement proposal to MoAI-ADK

Command Relationship Map

flowchart TD
    A[Utility commands] --> B[Automation commands]
    A --> C[Code management commands]
    A --> D[Feedback commands]

    B --> E["/moai
fully autonomous automation"]
    B --> F["/moai loop
iterative fix loop"]
    B --> G["/moai fix
one-shot auto-fix"]
    C --> H["/moai clean
dead-code removal"]
    C --> I["/moai codemaps
architecture doc generation"]
    C --> K["/moai gate
pre-commit quality gate"]
    C --> L["/moai mx
@MX code annotations"]
    C --> M["/moai review
multi-lens code review"]
    C --> N["/moai e2e
E2E testing"]
    D --> J["/moai feedback
submit feedback"]

    E -->|used internally| F
    F -->|run just once| G
    H -->|regenerate after cleanup| I

Info
For the SPEC 3-Phase lifecycle commands (/moai plan, run, sync, project, design, harness), see the workflow commands section. The full list of 16 subcommands the /moai orchestrator routes is also there.

Info
Not sure which command to use?
Want to build a whole feature at once → /moai
“Work on your own until this condition is met” → /moai goal
Lots of errors in the code you want fixed iteratively → /moai loop
Just want simple lint errors fixed fast → /moai fix
Just want a fast pass/fail check before committing → /moai gate
Want to plant AI-context code annotations (@MX) → /moai mx
Want to clean up unused code → /moai clean
Want to produce project architecture docs → /moai codemaps
A problem with MoAI-ADK itself → /moai feedback