create-nonoise

An AI SDLC bootstrapper. One command turns an empty directory into a fully-wired, AI-ready project — skills, docs hierarchy, cross-tool context files, and an advisor (Polly) that tells your team where they are in the SDLC and what to do next.

🌐 Site: https://nonoise-frmk.com 📦 Repo: https://github.com/russosalv/NONoise 📚 Docs: https://github.com/russosalv/NONoise/tree/main/docs

Quickstart

npx create-nonoise my-project
cd my-project
# Open in Claude Code, or tell Copilot: "start polly"

GitHub Release fallback. If the npm registry is temporarily unavailable, install the identical tarball from the GitHub Release:

npx https://github.com/russosalv/NONoise/releases/download/v1.6.0/create-nonoise-1.6.0.tgz my-project

The CLI asks which AI tools your team uses (Claude Code, GitHub Copilot), then scaffolds:

• src/ — your code. Stack-agnostic: pick .NET, Node, Python, Rust, Go.
• docs/ — six-folder source-of-truth hierarchy (architecture, requirements, calls, support, prd, sprints).
• .claude/skills/ — 40+ AI skills: Polly advisor, BMAD personas, Quint FPF validator, vendored superpowers, design and ops packs.
• Context files for each selected tool — CLAUDE.md, .github/copilot-instructions.md, AGENTS.md — generated from one source of truth.
• tools/md-extractor/ and tools/devops-push/ — Node CLIs.
• .nonoise/sdlc-flow.md — the SDLC flow Polly reads to detect where you are and suggest the next skill.

Options

npx create-nonoise [directory] [options]

  --workspace <kind>    new | existing-single | existing-multi
  --template <name>     single-project | multi-repo   (default: single-project)
  --ai <csv>            claude-code,copilot,codex,cursor,gemini-cli
  --user-name <name>    developer name (used by AI to address you)
  --user-locale <iso>   ISO 639-1 language code (en, it, …)
  --reverse / --no-reverse   enable/disable reverse-engineering config block
  --no-git              skip git init
  --yes, -y             non-interactive defaults
  --version, -v         print version
  --help, -h            print help

Maintenance (existing projects):
  --upgrade [path]        refresh bundled skills + re-install graphify
  --graphify-only [path]  narrow path: only the graphify CLI integration

When run with no flags, the CLI asks interactively whether to:

1. Create a new NONoise project (full scaffold)
2. Upgrade an existing project (refresh skills + graphify)
3. Force-install graphify only (narrow, idempotent fix)

If you pass a positional path that already contains nonoise.config.json, the CLI auto-detects it as an existing project and prompts: Upgrade / Graphify-only / Cancel — it never silently scaffolds over an existing project.

Upgrading an existing project

Bundled skills are copied at scaffold time and stay frozen at that version — npm install -g create-nonoise doesn't reach them. To pick up improvements (e.g. hardened anti-pattern guards on reverse-engineering, new skills, the /index slash command), run:

# Interactive (auto-detect)
npx create-nonoise@latest path/to/existing-project

# Direct, non-interactive
npx create-nonoise@latest --upgrade path/to/existing-project

What --upgrade does:

• Re-copies all bundled skills under <project>/.claude/skills/ (overwrites — picks up the latest SKILL.md files).
• Re-runs the graphify install (graphify claude install / graphify copilot install) so <project>/CLAUDE.md and .claude/settings.json get the current hook.
• Does NOT touch templates — CLAUDE.md, AGENTS.md, copilot-instructions.md keep your edits.
• Does NOT touch nonoise.config.json.

For the narrower repair path (only graphify, no skill refresh):

npx create-nonoise@latest --graphify-only path/to/existing-project

In --yes mode auto-detection refuses to operate on an existing project and prints the explicit --upgrade / --graphify-only command you should run — non-interactive scripts cannot silently clobber an existing project.

The SDLC flow

Requirements → Discovery → Architecture → Sprint → Implementation → Acceptance → Operations

Every phase is annotated [pair] (multiple seniors, large model) or [solo] (one dev per task, smaller model, parallelisable). Invoke /polly to find out where you are and get the exact prompt for the next skill.

Knowledge graph (graphify)

Every scaffolded project ships an integration with graphify for AST + semantic + community-detection knowledge graphs.

No API key needed. The scaffold writes per-platform graphify skill files into the project itself — .claude/skills/graphify/SKILL.md for Claude Code, .copilot/skills/graphify/SKILL.md for Copilot, .agents/skills/graphify/SKILL.md for Codex. Anyone who clones the repo (or runs the CLI without Python/uv on PATH) immediately gets the subagent-dispatch flow — no ANTHROPIC_API_KEY prompt, no missing-skill fallback. The companion graphify <X> install is still run as belt-and-suspenders for the user-global install.

In Claude Code there is also a project-level /index slash command that wraps the /graphify skill — it builds the graph using your IDE's own model:

/index .                 # index the current directory
/index <path>            # index a specific path
/index . --update        # incremental — re-extract only changed files

Don't use the bare graphify extract CLI for indexing — it requires an external LLM API key. The /graphify slash skill (and the /index shortcut) runs the same pipeline through your IDE assistant's session.

Who this is for

Teams adopting AI-assisted development with Claude Code or GitHub Copilot as first-class AI tools (v1 scope). Cursor, Gemini CLI, and Codex work in best-effort mode — the skills are plain Markdown and any tool can follow them.

Learn more

• Overview & philosophy
• Full SDLC flow
• Polly decision tree
• Skill catalog (40+ skills)
• Cross-tool support matrix

License

MIT, with attribution. See the repo for full LICENSE and ATTRIBUTION.md.

Created by Alessandro Russo.