MoiCle

Reusable AI agents, commands, skills, and architecture references for Claude Code

If you find this project useful, please consider giving it a ⭐ on GitHub!

A toolkit to bootstrap and accelerate project development with Claude Code through specialized agents, automation commands, workflow skills, and unified architecture references.

Features

• 18 AI Agents - 6 developer agents + 12 utility agents
• 2 Commands - Wizards for project bootstrap and go-to-market planning
• 10 Skills - Mode-based, auto-triggered workflows for the full SDLC (feature, bug, review, research, docs, marketing) plus standalone /challenge (self-critique) and /cleanup (dead-code/duplicate)
• 9 Architecture References - DDD + Hexagonal + stack-specific patterns

Current Support

• [x] Claude
• [x] Codex CLI
• [x] Antigravity
• [x] Cursor
• [x] Devin CLI
• [x] Windsurf

Older MoiCle versions merged agents into ~/.cursor/AGENTS.md. Re-run moicle install --target cursor for native .cursor/rules/*.mdc layout; you may delete legacy AGENTS.md manually.

Upgrading & stale items

moicle upgrade bumps the npm package only — items already installed in ~/.claude/ (or ./.claude/, .cursor/…) stay untouched. When a MoiCle release removes an asset (e.g. a skill or command deleted from the package), the leftover files remain active in your editor until removed.

Starting with this release, moicle install writes a .moicle-manifest.json into the target dir recording exactly what it placed. moicle uninstall removes everything in that manifest — including items the newer package no longer ships — while leaving your own files alone:

# Clean upgrade path
moicle uninstall && moicle install

Leftovers from pre-manifest versions (< this release) aren't tracked — remove them manually if needed, e.g. ~/.claude/commands/doc.md, ~/.claude/skills/feature-track/.

Installation

npm install -g moicle

Quick Start

# Install agents, commands, skills, architecture
moicle install

# Install for Codex CLI
moicle install --target codex --global

# Install for Antigravity
moicle install --target antigravity --global

# Install for Cursor (rules, commands, skills, architecture)
moicle install --target cursor --global
moicle install --target cursor --project

# Choose:
# 1. Pick Claude Code, Codex CLI, or Antigravity
# 2. Pick global or project scope

CLI Commands

| Command | Description | |---------|-------------| | moicle install | Interactive installation menu | | moicle install --global | Install to ~/.claude/ (symlinks) | | moicle install --project | Install to ./.claude/ (copies) | | moicle install --target codex --global | Install Codex skills + architecture to ~/.codex/ | | moicle install --target codex --project | Install Codex skills + architecture to ./.codex/ | | moicle install --target cursor --global | Install Cursor rules, commands, skills to ~/.cursor/ | | moicle install --target cursor --project | Install Cursor assets to ./.cursor/ | | moicle list --target cursor | List Cursor rules, commands, and skills | | moicle status | Show enabled/disabled status | | moicle enable <item> | Enable an agent/command/skill | | moicle disable <item> | Disable an agent/command/skill | | moicle uninstall | Remove installations |

What's Included

Architecture References (9)

| File | Description | |------|-------------| | clean-architecture.md | Core Clean Architecture principles | | hexagonal-architecture.md | Ports & Adapters boundary pattern | | go-backend.md | Go + Gin project structure | | laravel-backend.md | Laravel + PHP project structure | | nodejs-nestjs.md | Node.js + NestJS + Prisma (DDD + Hexagonal) | | react-frontend.md | React + Vite project structure | | remix-fullstack.md | Remix fullstack structure | | flutter-mobile.md | Flutter mobile structure | | monorepo.md | Monorepo structure |

Developer Agents (6)

| Agent | Description | |-------|-------------| | @flutter-mobile-dev | Flutter/Dart mobile & desktop development | | @go-backend-dev | Go + Gin backend API development | | @laravel-backend-dev | Laravel + PHP backend API development | | @nodejs-backend-dev | Node.js + NestJS + Prisma backend development | | @react-frontend-dev | React + TypeScript frontend development | | @remix-fullstack-dev | Remix full-stack development |

Utility Agents (12)

| Agent | Description | |-------|-------------| | @api-designer | RESTful & GraphQL API design | | @brainstormer | Structured ideation — 6 proven frameworks | | @clean-architect | Clean Architecture + MVVM patterns | | @code-reviewer | Code review for quality, security, performance | | @db-designer | Database schema design | | @devops | CI/CD, Docker, Kubernetes, Infrastructure | | @docs-writer | Technical documentation | | @perf-optimizer | Performance analysis & optimization | | @refactor | Code refactoring & cleanup | | @researcher | Delegated exploration — prior art, libraries, trade-offs | | @security-audit | Security vulnerability analysis | | @test-writer | Unit & integration test writing |

Commands (2)

| Command | Description | |---------|-------------| | /bootstrap | Wizard to create new project with 5 stack options | | /marketing | Unified go-to-market plan (brand + video + content) |

Skills (10)

Skills are grouped by a <group>- prefix. Type /<group>- then Tab in Claude Code to see all skills in a group. Each skill bundles several related workflows into modes chosen at the top of the file — pick a mode (or let Claude pick it from your natural-language phrasing). Two skills — /challenge and /cleanup — are intentionally standalone (no group prefix).

/feature-* — Build & Change

| Skill | Modes | When to use | |-------|-------|-------------| | /feature-build | NEW · REFACTOR · API · DEPRECATE · TRACK | Build a new DDD feature, restructure existing code, add/integrate an API, safely sunset a feature, or run a tracked checklist loop (branch → per-item test/implement/verify/commit) |

/fix-* — Bugs & Incidents

| Skill | Modes | When to use | |-------|-------|-------------| | /fix-bug | QUICK · DEEP | QUICK: fix fast with rollback plan · DEEP: trace a hard / recurring / intermittent bug | | /fix-incident | — | Production outage / on-call workflow |

/review-* — Review & Quality

| Skill | Modes | When to use | |-------|-------|-------------| | /review-code | SELF · PR · ARCHITECT · TDD · ADDRESS | Self-review a branch, review a teammate's PR, deep DDD audit, test-first development, or address PR comments |

/research-* — Explore & Learn

| Skill | Modes | When to use | |-------|-------|-------------| | /research-explore | WEB · SPIKE · ONBOARDING | Research solutions via docs, validate by building a prototype, or ramp up on a new codebase |

/docs-* — Project Documentation

| Skill | Modes | When to use | |-------|-------|-------------| | /docs-sync | SINGLE · FULL | SINGLE: hand-author one doc · FULL: auto-generate the whole docs site with review loop |

/marketing-* — Brand & Content (wrapped by the /marketing command)

| Skill | Modes | When to use | |-------|-------|-------------| | /marketing-content | STRATEGY · POST | Multi-post content strategy, or write ONE evergreen post optimized for Search + AI | | /marketing-brand | LOGO · VIDEO | Logo + brand identity spec, or video script/storyboard/production plan |

/challenge — Self-Critique (standalone, no group)

| Skill | Modes | When to use | |-------|-------|-------------| | /challenge | — | A quick skeptical re-examination of what you just did (or selected code): simpler? duplicated? optimal? edge cases? — reports honest findings without self-congratulating. Lighter than /review-code SELF. |

/cleanup — Dead Code & Duplicate (standalone, no group)

| Skill | Modes | When to use | |-------|-------|-------------| | /cleanup | DEAD-CODE · DUPLICATE | Safely remove dead code / consolidate duplicates. Detects with stack tools, verifies each hit with grep, then classifies DELETE (true dead) vs KEEP (reusable utils/scopes/helpers — asks when unsure). Removes per-group with build+test each. |

Skill decision matrix

When more than one skill / mode could fit, use this matrix:

| Situation | Use | Not | |-----------|-----|-----| | Bug just happened in prod, need fix in <1h | /fix-bug (QUICK) | /fix-bug DEEP (too slow) | | Bug keeps coming back after "fixes" | /fix-bug (DEEP) | /fix-bug QUICK (will repeat) | | About to push / open PR | /review-code (SELF) | /review-code PR (that's for others') | | Reviewing teammate's PR | /review-code (PR) | /review-code SELF (that's for own branch) | | Want to verify DDD compliance only | /review-code (ARCHITECT) | /review-code PR (broader scope) | | Don't know the right solution yet | /research-explore (WEB) | SPIKE (skip if you can decide from docs) | | Need to validate an idea by building | /research-explore (SPIKE) | /feature-build NEW (commit only after spike) | | Driving a multi-step task as a checklist with commit-per-step | /feature-build (TRACK) | /feature-build NEW (single full DDD feature), /review-code TDD (one unit, no branch/commit) | | Writing README / API docs by hand | /docs-sync (SINGLE) | /docs-sync FULL (overkill for single file) | | Generating full docs site from codebase | /docs-sync (FULL) | /docs-sync SINGLE (manual is slower) | | Quick "did I overcomplicate this?" gut-check on fresh code | /challenge | /review-code ARCHITECT (heavier, scored audit) |

Backward compatibility

The 22 original skills were consolidated into 10 mode-based skills. Old trigger phrases still work — they're kept in each skill's description so Claude auto-invokes the right skill and the right mode when the user says e.g. "deep debug" (→ /fix-bug DEEP), "hotfix" (→ /fix-bug QUICK), "review changes" (→ /review-code SELF). The flattened name /group-action is the explicit invocation form.

Architecture-First Approach

All agents reference architecture files to ensure consistency:

~/.claude/
├── agents/
├── commands/
├── skills/
└── architecture/          # Architecture references
    ├── ddd-architecture.md
    ├── flutter-mobile.md
    ├── go-backend.md
    ├── hexagonal-architecture.md
    ├── laravel-backend.md
    ├── monorepo.md
    ├── nodejs-nestjs.md
    ├── react-frontend.md
    ├── remix-fullstack.md
    └── _shared/           # read-project-first, engineering-principles, stack-detection, severity-levels

When an agent is invoked, it reads the architecture file first before coding according to the defined structure.

For Codex CLI, MoiCle installs architecture docs into ~/.codex/architecture or ./.codex/architecture, and converts MoiCle agents, commands, and existing skills into native Codex skills under .codex/skills. Restart Codex after a global install so the new skills are loaded.

For Antigravity, MoiCle installs architecture docs into ~/.gemini/architecture or ./.gemini/architecture, and converts MoiCle agents, commands, and existing skills into native Antigravity skills under .gemini/skills.

For Devin CLI, MoiCle installs agents as native subagent profiles into ~/.config/devin/agents or ./.devin/agents, converts commands and skills into native Devin skills under ~/.config/devin/skills or ./.devin/skills, and installs architecture docs into <root>/architecture. Ask Devin to use a profile by name (e.g. "review with the code-reviewer subagent") or invoke skills with /skill-name.

Usage Examples

Using Commands

# Bootstrap new project
/bootstrap                              # Interactive wizard
/bootstrap go                           # Quick start with Go + Gin
/bootstrap react                        # Quick start with React + Vite
/bootstrap flutter                      # Quick start with Flutter
/bootstrap landing using monorepo       # Create landing page with monorepo structure

# Go-to-market plan
/marketing                              # Unified brand + video + content plan

Using Agents

# Agents will automatically read architecture reference before coding
@go-backend-dev Implement user authentication module
@react-frontend-dev Create dashboard page with charts
@clean-architect Review this PR's architecture

Enable/Disable Agents

# Disable unused agents
moicle disable @rust-dev
moicle disable @swift-ios-dev

# Enable again
moicle enable @rust-dev

# View status
moicle status

Project Structure

claude-agents-kit/
├── bin/
│   └── cli.js
├── src/
│   ├── commands/
│   │   ├── install.js
│   │   ├── uninstall.js
│   │   ├── list.js
│   │   ├── enable.js
│   │   ├── disable.js
│   │   └── status.js
│   └── utils/
│       ├── symlink.js
│       └── config.js
├── assets/
│   ├── agents/
│   │   ├── developers/
│   │   └── utilities/
│   ├── architecture/        # Architecture references
│   ├── commands/
│   └── skills/
├── package.json
└── README.md

Requirements

• Node.js >= 18.0.0
• Claude Code CLI

Donate

If you find this project useful, consider buying me a coffee!

USDT / ETH / BNB:

0xf3b2a531cb2774c77a751cdb10e043992eff5a2c

| Network | Token | |---------|-------| | Ethereum (ERC-20) | USDT, ETH | | Binance Smart Chain (BEP-20) | USDT, BNB |

License

GPL-3.0