LAZY_DEV Framework
Version: 2.2.2 | License: MIT | Status: Production-Ready | Platforms: Linux, macOS, Windows
Quick Start
3-minute setup to first working feature:
# 1. Copy LAZY_DEV to your project
cp -r LAZY_DEV/.claude/ .claude/
# 2. Set required environment variable
export ENRICHMENT_MODEL=claude-3-5-haiku
# 3. Plan your first feature
/lazy plan "Add user authentication"
# 4. Execute first task
/lazy code TASK-1
# 5. Review and create PR
/lazy review US-3.4
What just happened? PM agent created user stories, Coder agent implemented with TDD, quality pipeline validated code, PR was auto-created.
What & Why
The Problem
Modern AI-assisted development suffers from:
- Inconsistent quality (no format/lint/type/test enforcement)
- Lost context between sessions
- Manual repetitive tasks (stories, commits, reviews)
- Fragmented tools scattered across projects
- Constant re-prompting of the same context
The Solution
LAZY_DEV provides a turnkey framework with:
- Automatic quality pipeline: format → lint → type → test (fail-fast)
- Persistent knowledge: MCP Memory graph across sessions (project-isolated)
- End-to-end automation: voice → feature → tasks → code → PR
- Protective hooks: prevent dangerous operations (rm -rf, force-push to main)
- Specialized agents: intelligent delegation based on context
Philosophy
"Be lazy, but consistently excellent"
Automate the mundane (formatting, commits, PRs) while enforcing discipline (quality gates, TDD, security checks).
Core Features
8 Commands
/lazy plan <brief>- Generate US-story + tasks/lazy code <input>- Implement from story/task/brief/issue/lazy review <id>- Review feature, create PR or generate debug report/lazy fix <report>- Apply review fixes from debug report/lazy docs <file>- Generate/update docs/lazy clean <path>- Remove dead code/lazy memory-graph <statement>- Persist to MCP/lazy memory-check- Verify MCP connectivity
9 Specialized Agents (automatic delegation)
- project-manager, coder, reviewer, reviewer-story, tester
- research, documentation, refactor, cleanup
17 Skills (reusable patterns)
- Planning: story-traceability, task-slicer, ac-expander
- Development: test-driven-development, diff-scope-minimizer
- Quality: code-review-checklist, security-scanner
- Documentation: readme-template, api-doc-generator
- Integration: gh-issue-sync, mcp-memory-router
4 Hook Types (10 implementations)
- UserPromptSubmit: context enrichment, memory detection, style selection
- PreToolUse: safety checks, audit logging, validation
- PostToolUse: auto-formatting, memory suggestions, cleanup
- Stop: quality gate logging, TDD enforcement, metrics
Installation
Cross-Platform Support ✅
LAZY_DEV works seamlessly on Linux, macOS, and Windows. All hooks, scripts, and tools are cross-platform by design.
Platform-specific guides:
- All Platforms: CROSS_PLATFORM.md - Complete compatibility guide
- Windows Users: WINDOWS_SETUP.md - Windows-specific setup
Prerequisites
- Python 3.11+
- Claude Code CLI (latest)
- uv package manager (
pip install uv) - git
- gh CLI (for PR creation)
- Node.js (for MCP Memory, optional)
- Windows only: Git Bash or WSL (for bash scripts)
Step 1: Install Framework
# Copy LAZY_DEV to your project
cp -r LAZY_DEV/.claude/ .claude/
Or via plugin marketplace:
# Add marketplace
/plugin marketplace add MacroMan5/claude-code-workflow-plugins
# Install plugin
/plugin install lazy-dev@MacroMan5
# Restart Claude Code
Step 2: Set Environment Variables
# Required for pre-prompt enrichment
export ENRICHMENT_MODEL=claude-3-5-haiku
# Optional: Enable strict mode
export LAZYDEV_ENFORCE_TDD=1
export LAZYDEV_MIN_TESTS=3