sdlc-wizard-cowork

Claude Code SDLC Wizard

A self-evolving Software Development Life Cycle (SDLC) enforcement system for AI coding agents. Makes Claude plan before coding, test before shipping, and escalate when uncertain. Measures itself getting better over time.

Built on 15+ years of software engineering and founding engineering experience — battle-tested patterns from real production systems, baked into an AI agent that follows tried-and-true software quality practices so you don't have to enforce them manually.

Built for Claude Code. Using OpenAI's Codex CLI instead? Check out codex-sdlc-wizard. Need privacy-first / any-backend (local Ollama, Azure OpenAI, hosted OSS)? See opencode-sdlc-wizard. (Full ecosystem.)

Install

Requires Claude Code (Anthropic's CLI for Claude).

Run from your terminal or from inside Claude Code (! prefix):

npx -y agentic-sdlc-wizard@latest init

The @latest pin forces npm to fetch the newest version. Without it, npx may serve a stale CLI from your local cache (#358); init also nudges if it detects a gap. Then start (or restart) Claude Code — type /exit then claude to reload hooks. Setup auto-invokes on first prompt — Claude reads the wizard doc, scans your project, and generates bespoke CLAUDE.md, SDLC.md, TESTING.md, and ARCHITECTURE.md. No manual commands needed.

curl -fsSL https://raw.githubusercontent.com/BaseInfinity/claude-sdlc-wizard/main/install.sh | bash

brew install BaseInfinity/sdlc-wizard/sdlc-wizard
sdlc-wizard init

gh extension install BaseInfinity/gh-sdlc-wizard
gh sdlc-wizard init

npx github:BaseInfinity/claude-sdlc-wizard init

npm install -g agentic-sdlc-wizard
sdlc-wizard init

npx agentic-sdlc-wizard check        # Human-readable
npx agentic-sdlc-wizard check --json  # Machine-readable (CI-friendly)

Why Use This

You want Claude Code to follow engineering discipline automatically:

• Plan before coding (not guess-and-check)
• Write tests first (TDD enforced via hooks)
• State confidence (LOW = escalate to a model first, don't guess)
• Track work visibly (TaskCreate)
• Self-review before presenting
• Prove it's better (use native features unless you prove custom wins)

The wizard auto-detects your stack (package.json, test framework, deployment targets) and generates bespoke hooks + skills + docs. CI validates the generated assets; cross-stack setup-path E2E is on the roadmap.

What This Actually Is

Five layers working together:

Layer 5: SELF-IMPROVEMENT
  Weekly/monthly workflows detect changes, test them
  statistically, create PRs. Baselines evolve organically.

Layer 4: STATISTICAL VALIDATION
  E2E scoring with 95% CI (5 trials, t-distribution).
  SDP normalizes for model quality. CUSUM catches drift.

Layer 3: SCORING ENGINE
  Multi-criteria scoring, 10/11 points. Claude evaluates Claude.
  Before/after wizard A/B comparison in CI.

Layer 2: ENFORCEMENT
  Hooks fire every interaction (~100 tokens).
  PreToolUse reminds Claude to write tests first.

Layer 1: PHILOSOPHY
  The wizard document. KISS. TDD. Confidence levels.
  Copy it, run setup, get a bespoke SDLC.

What Makes This Different