Lorenzo's Claude Code Plugin

Minimal Claude Code plugin for Next.js + React + Supabase development.

18 commands | 6 agents | 10 skills | 9 hooks | 4 MCP servers

---

Why This Plugin?

Most Claude Code plugins are bloated. This one is focused:

Feature	This Plugin	Others
Commands	18 (focused)	30-66 (bloated)
MCP Servers	4 (minimal context)	8-20+ (context hog)
Circuit Breaker	Detects thrashing loops	Usually missing
Verification Gates	Evidence-based completion	Trust-based
Micro-Tasking	2-5 min task enforcement	No enforcement
Target Stack	Next.js + React + Supabase	Everything (unfocused)

Innovative Features

Circuit Breaker - Detects when Claude is stuck in a loop:

• Same error 3+ times? Stops and suggests pivot
• Same file edited 3+ times? Flags scope creep
• Extracts error signatures to avoid false negatives

Verification First - No "it should work" claims:

• 5-step gate: IDENTIFY → RUN → READ → VERIFY → CLAIM
• Requires actual command output, not mental simulation
• Blocks hedging language ("probably", "seems to")

Micro-Tasking - Prevents scope creep:

• Tasks must be 2-5 minutes
• Decomposition templates for features, bugs, refactors
• If it takes >10 min, it wasn't properly decomposed

---

Quick Start

Option 1: npm (Recommended)

# Install globally
npm install -g @gr8monk3ys/claude-code-plugin

# Run installer
lorenzo-claude install

# Verify installation
lorenzo-claude doctor

Option 2: npx (No Install)

npx @gr8monk3ys/claude-code-plugin install

Option 3: Plugin Marketplace

# In Claude Code
/plugin marketplace add https://github.com/gr8monk3ys/lorenzos-claude-code
/plugin install lorenzos-claude-code

Option 4: Manual

git clone https://github.com/gr8monk3ys/lorenzos-claude-code.git
cp -r lorenzos-claude-code/.claude/* ~/.claude/

---

Commands

UI & Components

Command	Description
/component-new	Create React component with TypeScript
/page-new	Create Next.js page with App Router
/hook-new	Create custom React hook

API & Backend

Command	Description
/api-new	Create API endpoint with Zod validation
/api-test	Generate API tests

Testing & Quality

Command	Description
/test-new	Generate test files
/verify	Run 6-phase quality gates before PR
/lint	Run linting
/review	PR review checklist

Supabase

Command	Description
/types-gen	Generate TypeScript types from database
/edge-function-new	Create Supabase Edge Function

Workflow

Command	Description
/plan	Create implementation plan
/deploy	Generate deployment config
/handoff	Create session handoff document
/memory	View/update persistent memory
/learn	Extract patterns from session
/evolve	Evolve instincts into skills
/research	Deep topic exploration

---

Agents

Agent	Use Case
code-reviewer	PR reviews, security, quality
backend-architect	API design, database architecture
frontend-architect	React/Next.js patterns, accessibility
build-error-resolver	Fix TypeScript and build errors (<5% diff)
test-strategist	Test planning and coverage
devops-engineer	CI/CD, deployment

---

Skills

Auto-activate based on context:

Skill	Purpose	Priority
circuit-breaker	Prevent infinite loops	95
verification-first	Quality gates before completion	90
micro-tasking	Break work into small tasks	-
api-development	Next.js API patterns	-
frontend-development	React best practices	-
database-operations	Supabase patterns	-
memory-persistence	Session context preservation	-
code-quality	Code review patterns	-
continuous-learning	Pattern extraction & evolution	-
research	Structured exploration workflow	-

---

Hooks

Hook	Event	Purpose
session-start	SessionStart	Restore previous context
session-end	Stop	Save session state
block-sensitive-files	PreToolUse	Prevent editing secrets
validate-json	PreToolUse	Validate JSON before write
auto-format	PostToolUse	Format code after edits
circuit-breaker	PostToolUse	Detect stuck loops
pre-compact	PreCompact	Save context before compact
notify-completion	Stop	Desktop notification
skill-activator	UserPromptSubmit	Auto-evaluate & suggest skills

---

Skill Auto-Evaluation (NEW)

Unlike other plugins that use simple keyword matching, this plugin analyzes prompts across 5 dimensions with weighted confidence scoring:

Dimension	Weight	Example
Keywords	2pts	"api", "component", "test"
Patterns	3pts	route.ts, use[A-Z].*()
File paths	4pts	app/api/users/route.ts
Directories	5pts	prisma/, components/
Intents	4pts	"create.*api", "fix.*bug"

Thresholds:

• ≥8 points: Skill auto-activated
• ≥5 points: Skill suggested
• <5 points: No recommendation

Example output:

Prompt: "Fix the bug in app/api/users/route.ts - keeps failing"

<skill-evaluation>
  <skill name="circuit-breaker" confidence="24" status="activate">
    <matches>
      keyword:failing (+2pts)
      keyword:keeps (+2pts)
      pattern:same\s+error (+3pts)
      ...
    </matches>
  </skill>
  <skill name="api-development" confidence="24" status="activate">
    <matches>
      filepath:app/api/users/route.ts (+4pts)
      directory:app/api (+5pts)
      ...
    </matches>
  </skill>
</skill-evaluation>

Configure rules in .claude/hooks/skill-rules.json.

---

MCP Servers

Only 4 essential servers (CLI-first approach):

Server	Purpose
context7	Library documentation
memory	Cross-session memory
playwright	Browser automation
github	PRs and issues

---

CLI Commands

# Install plugin to ~/.claude/
lorenzo-claude install

# Update existing installation
lorenzo-claude update

# Verify installation health
lorenzo-claude doctor

# Remove plugin
lorenzo-claude uninstall

# Show version
lorenzo-claude version

# Show help
lorenzo-claude help

Short alias: lcc (e.g., lcc install)

---

Comparison to Alternatives

Feature	This Plugin	everything-claude-code	SuperClaude
Focus	Next.js/React/Supabase	General purpose	General purpose
Commands	18	20+	30
Agents	6	15+	16
MCP Servers	4	8+	8
Circuit Breaker	Yes (innovative)	Basic	No
Verification Gates	Yes (5-step)	Basic	No
Micro-Tasking	Yes (2-5 min)	No	No
Install Method	npm, plugin, manual	Plugin, manual	pip, npm
Context Overhead	Low	Medium	High

Choose this plugin if:

• You work primarily with Next.js + React + Supabase
• You want guardrails against AI failure modes
• You prefer minimal context overhead

Choose alternatives if:

• You need multi-language support (Go, Rust, Python)
• You want deep research workflows
• You prefer maximum features over focus

---

Requirements

• Claude Code 2.0.13+
• Node.js 18+

---

Contributing

Contributions welcome! See CONTRIBUTING.md.

---

License

GPL-3.0

---

Star History