sdlc

SDLC Plugin for Claude Code

A comprehensive Claude Code plugin that enhances your software development lifecycle with specialized agents, commands, and integrations.

Features

🤖 Specialized Agents

• codebase-analyzer - Deep implementation analysis with precise file:line references
• codebase-locator - Fast component location and discovery across your codebase
• codebase-pattern-finder - Pattern detection and architectural understanding
• web-search-researcher - Real-time research capabilities via web search
• implementer - Fresh-context task implementer for subagent workflow (TDD-aware, self-reviews)
• spec-reviewer - Verifies implementation matches spec exactly (nothing missing, nothing extra)
• code-quality-reviewer - Quick sanity check for obvious bugs and code smells
• test-writer - Dispatched delegate paired with the test skill; writes tests following Kent C. Dodds shape rules (flat structure, factories, disposables)
• research-synthesizer - Merges multiple research reports into comprehensive document with theme-based synthesis; surfaces consensus findings, disagreements, and novel insights from cross-pollination

⚡ Commands

• /plan - Generate detailed implementation plans with phases, complexity analysis, and validation steps
• /research - AI-powered research with project context
• /research-deep - Parallel research with 3 instances and synthesis for comprehensive coverage
• /implement - Execute implementation based on generated plans
• /review - Parallel code review with Codex and Gemini (models resolved from config/model-registry.md), consolidated findings
• /submit - Prepare and submit work for review
• /verify - Validate implementation against acceptance criteria

🎯 Skills

• codex - OpenAI Codex integration for code analysis, review, refactoring, and automated editing (model auto-resolved from registry)
• gemini - Google Gemini integration for code review, plan analysis, and big context (>200k) processing (model auto-resolved from registry)
• update-models - Re-resolve the model registry by querying OpenAI, Google AI, and Oracle CLI for latest models
• interview - Deep interviews about any topic with iterative questioning (uses Opus)
• tdd - TDD enforcement during implementation (strict/soft/off modes via CLAUDE.md)
• test - Write or review tests following Kent C. Dodds principles (flat structure, composable setup, disposable fixtures); paired with the test-writer agent
• domain-model - Grill plan language against the target repo's CONTEXT.md; surface vocabulary conflicts, lazily create CONTEXT.md, and offer ADRs when architectural decisions meet the 3-criteria bar
• finish-branch - Post-merge cleanup: switches to main, pulls, runs tests, removes worktree
• agent-change-walkthrough - Narrative walkthrough of AI-authored code changes after implementation; explains what changed, why, and how it behaves
• x-search - Search X/Twitter for real-time developer discourse, product feedback, community sentiment, and expert opinions
• judgment-eval - Evaluates agent judgment quality through scenario-based testing in-conversation
• constitution-compliance-review - Scores plugin commands, agents, and skills against Anthropic Constitutional AI principles; identifies rule-based vs reasoning-based spectrum improvements
• system-prompt-clinic - Interactive diagnostic that transforms rule-based system prompts to reasoning-based using Constitutional AI principles

🔌 Integrations

• Context7 - Up-to-date library documentation and code examples
• Perplexity Sonar (via OpenRouter) - Advanced web search and research capabilities

Installation

Prerequisites

• Claude Code CLI
• Node.js 18+ (for MCP servers)
• Bun (recommended) or npm

Install Plugin

# Clone or download the plugin
git clone <repository-url> sdlc-plugin

# Navigate to your Claude Code plugins directory
cd ~/.claude/plugins

# Copy or symlink the plugin
ln -s /path/to/sdlc-plugin sdlc-plugin

# Verify installation
claude --debug

Environment Setup

Create a .env file or set environment variables:

# Required for Perplexity Sonar integration (via OpenRouter)
export OPENROUTER_API_KEY="your-api-key"
export OPENROUTER_MODEL="perplexity/sonar-pro"  # or any OpenRouter model

Usage

Complete Workflow: Research → Plan → Issue → Implementation → PR

This plugin implements a comprehensive SDLC workflow with proper separation of concerns:

Research (Documentation)      → Plan (Specification)     → Issue (Execution)     → Code (Implementation)
research/*.md (committed)       plans/*.md (committed)     #123 (GitHub)          Branch + PR

1. Research Phase

/research "How does Next.js handle server-side rendering?"

Creates: research/[topic].md (committed to git)