moai

Orchestrates autonomous development workflows by routing tasks to specialized agents.

Updated Jun 18, 2026
One-click install
npx skills add https://github.com/h102-log/pdfrag --skill moai-h102-log
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: moai
Source: https://github.com/h102-log/pdfrag/tree/main/.claude/skills/moai
Command: npx skills add https://github.com/h102-log/pdfrag --skill moai-h102-log

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Coordinating multi-step software development work—planning specifications, implementing code, fixing errors, and syncing documentation—requires constant manual orchestration. This Skill acts as a unified orchestrator that routes natural language requests or explicit subcommands to specialized agents, automating the full plan-to-deployment pipeline. ## Core Features & Use Cases - Intent-Based Routing: Parses subcommands (plan, run, sync, fix, loop, project, feedback) or natural language to select the correct workflow, with automatic SPEC-ID detection. - Plan-Run-Sync Pipeline: Creates EARS-format SPEC documents, implements them via Domain-Driven Development with TRUST 5 quality validation, and synchronizes documentation with PR creation. - Autonomous Fixing: Detects and fixes LSP errors, lint issues, and test failures in one-shot or iterative loop modes with snapshot-based resume support. - Use Case: A developer types a feature request in natural language; the orchestrator explores the codebase, generates an approved SPEC, delegates implementation to expert agents, validates quality gates, and prepares a pull request. ## Quick Start Ask the agent to plan and implement a new user authentication feature end to end using the moai workflow.

Frequently Asked Questions about moai

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run an autonomous development workflow in Claude Code?▼

Invoke the moai skill with a natural language task or an explicit subcommand like plan, run, or sync. The orchestrator routes the request to specialized agents and executes the plan-run-sync pipeline with user approval checkpoints.

How to automatically fix lint and type errors in a project?▼

Use the fix subcommand for a one-shot parallel scan with LSP diagnostics, AST-grep, and linters, or the loop subcommand for iterative fixing until quality gates pass. Fixes are classified into four levels, with Level 1-2 applied automatically.

What is the EARS format used for SPEC documents?▼

EARS is the requirements format used by the plan workflow to generate specification documents. Each SPEC produces spec.md, plan.md, and acceptance.md files with YAML frontmatter, requirement modules, and Given/When/Then acceptance scenarios.

Can I resume an interrupted workflow session?▼

Yes, most workflows support a --resume flag with a SPEC-ID or snapshot identifier. Snapshots are stored in .moai/cache/ directories, letting you continue plan, run, fix, or loop workflows from the last completed checkpoint.

Does the orchestrator write code directly?▼

No, all implementation is delegated to specialized agents such as expert-backend, expert-frontend, and manager-ddd via Task calls. The orchestrator only routes intents, collects user approvals, and tracks task progress.