AI Software Factory

Turn Claude Code into a virtual engineering team. 133 skills. One person.

🚀 One-Command Install (any device)

Zero to Claude Code in a single paste. Pick your OS.

💻 Windows (PowerShell as Admin)

Set-ExecutionPolicy Bypass -Scope Process -Force; winget install OpenJS.NodeJS.LTS -e --accept-source-agreements --accept-package-agreements; winget install Git.Git -e --accept-source-agreements --accept-package-agreements; $env:Path = [System.Environment]::GetEnvironmentVariable("Path","Machine") + ";" + [System.Environment]::GetEnvironmentVariable("Path","User"); node --version; npm --version; npm install -g @anthropic-ai/claude-code; [System.Environment]::SetEnvironmentVariable("CLAUDE_CODE_GIT_BASH_PATH","C:\Program Files\Git\bin\bash.exe","User"); $env:CLAUDE_CODE_GIT_BASH_PATH="C:\Program Files\Git\bin\bash.exe"; claude

🍎 Mac (Terminal)

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)" && brew install node && node --version && npm --version && npm install -g @anthropic-ai/claude-code && claude

📱 Android (Termux)

pkg upgrade -y && pkg install nodejs-lts -y && node --version && npm --version && npm install -g @anthropic-ai/claude-code && claude

🔥 Optional — Repo link

https://github.com/francky180/franc-system-template

⚡ Optional — Clone + open

git clone https://github.com/francky180/franc-system-template && cd franc-system-template && claude

---

🤖 Auto-Rebuild Prompt — paste this into Claude after login

The install block above gets Claude Code running. This prompt makes Claude do the rest — clone the template, install all 114 skills + gstack + superpowers + the 47 agency-agents + 15 Seedance skills + BRAIN + auto-sync hook, list the MCPs to add, verify everything, and report back. You never leave Claude.

🚀 REBUILD MISSION — install the full AI Software Factory from inside Claude. Execute every step, don't ask, report at the end.

STEP 1 · CLONE
   git clone https://github.com/francky180/franc-system-template ~/franc-system-template

STEP 2 · INSTALL CORE (114 skills + gstack + superpowers + BRAIN + CLAUDE.md + settings + auto-sync hook)
   cd ~/franc-system-template && bash ai-tools/install-all.sh

STEP 3 · INSTALL AGENCY-AGENTS (47 specialist subagents — paid media, sales, finance, support, testing, marketing)
   git clone https://github.com/msitarzewski/agency-agents /tmp/agency-agents
   mkdir -p ~/.claude/agents
   cp /tmp/agency-agents/agents/*.md ~/.claude/agents/

STEP 4 · INSTALL SEEDANCE 2.0 SKILLS (15 Higgsfield prompt skills)
   git clone https://github.com/beshuaxian/higgsfield-seedance2-jineng /tmp/seedance
   for d in /tmp/seedance/skills/*/; do
     name=$(grep -m1 "^name:" "$d/SKILL.md" | sed "s/name: //")
     [ -d "$HOME/.claude/skills/$name" ] || cp -r "$d" "$HOME/.claude/skills/$name"
   done

STEP 5 · MCP SERVERS — add after login with `claude mcp add <name>` (your own API keys):
   gitnexus, vercel, playwright, stripe, codex, filesystem, adadvisor,
   tradingview, transcript-api, context7, sequential-thinking, shadcn-ui

STEP 6 · VERIFY — run each, capture output
   - node --version
   - git --version
   - claude --version
   - find ~/.claude/skills -maxdepth 1 -type d | wc -l    (expect 130+)
   - ls ~/.claude/agents/*.md | wc -l                      (expect 47+)
   - grep -c rebuild-sync ~/.claude/settings.json          (expect >= 1)

STEP 7 · REPORT
   ✅ Step 1 — cloned to ~/franc-system-template
   ✅ Step 2 — N skills · gstack · superpowers · BRAIN · auto-sync hook installed
   ✅ Step 3 — M agency-agents installed
   ✅ Step 4 — 15 Seedance prompt skills installed
   ✅ Step 5 — MCPs to add manually with claude mcp add
   ✅ Verify — Node vXX · Git X.X · Claude X.X · NNN skills · MM agents · hook present

   End with: "Rebuild complete. Type /help to see all skills, then /office-hours to start your first build."

RULES
• Execute, don't ask. No "can I?" — just do it.
• install-all.sh and the copies are additive (no-clobber) — they never delete your existing skills.
• If a step fails, don't halt. Complete the rest, then show the one-line fix for the failure.
• Tone: short, confident, no apologies, no filler.

What you end up with: 114 bundled skills + gstack (~35) + superpowers (~14) + 15 Seedance = ~178 skills, 47 agency-agents, the BRAIN vault, and the auto-sync hook. See Quick Start (5 minutes) below.

What You Get When You Clone This Repo

Bundled in this repo (114 skills — copied to ~/.claude/skills/ on install):

Category	Count	Examples
Build · Deploy · Ops	41	/ship /land-and-deploy /canary /investigate /qa /review /retro /health /cso /autoplan /plan-ceo-review /plan-eng-review
Creator Marketing	31	/creator-outreach-sequence-generator /campaign-roi-calculator /niche-fit-scorer /creator-rate-estimator /brand-safety-screen
Marketing Suite	15	/market-ads /market-funnel /market-launch /market-seo /market-social /market-copy /market-landing
Design System	15	/design-consultation /design-shotgun /banner-design /frontend-design /ui-styling /ui-ux-pro-max /slides
GitNexus (code graph)	8	/gitnexus-impact-analysis /gitnexus-debugging /gitnexus-refactoring /gitnexus-pr-review
Best Practices	4	/next-best-practices /vercel-react-best-practices /remotion-best-practices /playwright-best-practices

Plus installed automatically by install-all.sh:

• gstack (~35+ specialist skills) — virtual team clone: CEO mode, eng manager, design lead, release engineer
• Superpowers (~14 process skills) — Obra's process toolkit for disciplined AI workflows
• BRAIN vault at ~/BRAIN — Obsidian-navigable knowledge graph with the 100-book principle library
• CLAUDE.md global identity file with the 15 Always-On Principles + routing rules
• settings.json with the rebuild-sync PostToolUse hook (auto-snapshots every edit to a mirror repo)
• Claude Code CLI (if not already installed)
• Bun runtime (if not already installed)

Plus added by the Auto-Rebuild Prompt (Steps 3–4):

• 47 agency-agents — specialist subagents for paid media, sales, finance, support, testing, marketing (→ ~/.claude/agents/)
• 15 Seedance 2.0 skills — Higgsfield cinematic video prompt library

Post-install totals: 130+ skill folders in ~/.claude/skills/ (114 bundled + gstack + superpowers + 15 Seedance) and 47 agents in ~/.claude/agents/, plus the BRAIN vault and the auto-sync hook. Note: gstack is one folder that itself bundles a full ~290-skill engineering team, so your effective skill count is far higher. Add the 12 MCPs with claude mcp add to complete the stack.

Run /help inside Claude Code to list every installed skill.

The Sprint Lifecycle

Every task follows: Think > Plan > Build > Review > Test > Ship > Reflect

Phase	What Happens
Think	/office-hours asks 6 forcing questions before code
Plan	/autoplan runs CEO + design + eng review automatically
Build	Subagent-driven development with TDD enforced
Review	/review + /cso for code review + security audit
Test	/qa opens a real browser and tests your app
Ship	/ship runs tests, creates PR in one command
Reflect	/retro weekly retrospective, /learn compounds knowledge

Quick Start (5 minutes)

Zero to Claude Code in one command. Pick your OS and paste the block into your terminal — it installs Node.js, Claude Code, logs you in, and runs the install script.

All you need afterwards is a Claude Max subscription.

🪟 Windows (PowerShell — run as Administrator)

winget install OpenJS.NodeJS.LTS -e --accept-source-agreements --accept-package-agreements; `
winget install Git.Git -e --accept-source-agreements --accept-package-agreements; `
refreshenv; `
npm install -g @anthropic-ai/claude-code; `
git clone https://github.com/francky180/franc-system-template.git "$HOME\ai-software-factory"; `
cd "$HOME\ai-software-factory"; `
bash ai-tools/install-all.sh; `
claude

🍎 Mac (Terminal)

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)" && \
brew install node git && \
npm install -g @anthropic-ai/claude-code && \
git clone https://github.com/francky180/franc-system-template.git ~/ai-software-factory && \
cd ~/ai-software-factory && \
bash ai-tools/install-all.sh && \
claude

🐧 Linux (Debian / Ubuntu)

sudo apt update && sudo apt install -y curl git && \
curl -fsSL https://deb.nodesource.com/setup_lts.x | sudo -E bash - && \
sudo apt install -y nodejs && \
sudo npm install -g @anthropic-ai/claude-code && \
git clone https://github.com/francky180/franc-system-template.git ~/ai-software-factory && \
cd ~/ai-software-factory && \
bash ai-tools/install-all.sh && \
claude

📱 Termux (Android)

pkg upgrade -y && pkg install nodejs-lts git -y && \
npm install -g @anthropic-ai/claude-code && \
git clone https://github.com/francky180/franc-system-template.git ~/ai-software-factory && \
cd ~/ai-software-factory && \
bash ai-tools/install-all.sh && \
claude

What happens

1. Installs Node.js LTS + Git (if missing)
2. Installs Claude Code globally via npm
3. Clones this template to ~/ai-software-factory
4. Runs install-all.sh — copies all 115 skills into ~/.claude/skills/
5. Launches claude — opens browser for Claude Max login (first-time only)

Verify

Once Claude Code is running, type:

/help

You should see all 115 skills listed. If /office-hours, /ship, or /autoplan show up, you're done.

Prerequisites (what the scripts install for you)

• Git — installed automatically
• Node.js 18+ — installed automatically
• Claude Code — installed automatically
• Claude Max subscription — you need this separately (claude.ai/pricing)

All 115 Skills

gstack — Virtual Engineering Team (40+)

Command	Specialist	What They Do
/office-hours	YC Office Hours	6 forcing questions before code
/plan-ceo-review	CEO	Find the 10-star product
/plan-eng-review	Eng Manager	Lock architecture + test matrix
/plan-design-review	Designer	Rate design dimensions 0-10
/plan-devex-review	DX Lead	Developer experience audit
/autoplan	Pipeline	All reviews in one command
/review	Staff Engineer	Finds bugs CI misses
/cso	Security Officer	OWASP + STRIDE + secrets audit
/qa	QA Lead	Real browser testing + bug fixes
/qa-only	QA Reporter	Test and report, no fixes
/ship	Release Engineer	Tests + PR in one command
/land-and-deploy	Release Engineer	Merge + verify production
/deploy-to-vercel	Deploy	Push to Vercel
/canary	SRE	Post-deploy monitoring
/benchmark	Perf Engineer	Core Web Vitals regression detection
/investigate	Debugger	Root-cause debugging
/browse	QA Engineer	Headless Chromium browser
/open-gstack-browser	Browser	Visible AI-controlled Chrome
/connect-chrome	Browser	GStack Browser with sidebar
/webapp-testing	Tester	Playwright-based web app testing
/design-consultation	Design Lead	Design system + aesthetic planning
/design-shotgun	Design Explorer	AI mockup variants + comparison
/design-html	Design Engineer	Mockup to production HTML
/design-review	Visual QA	Find and fix visual issues
/retro	Eng Manager	Weekly retrospective with trends
/learn	Memory	Cross-session learnings
/health	Dashboard	Code quality 0-10 score
/codex	Second Opinion	Independent code review via Codex
/checkpoint	State	Save + resume working state
/document-release	Docs	Post-ship documentation update
/setup-deploy	Config	Configure deploy platform
/setup-browser-cookies	Auth	Import cookies for testing
/careful	Safety	Warns before destructive commands
/freeze	Safety	Lock edits to one directory
/unfreeze	Safety	Remove directory lock
/guard	Safety	Full safety mode (careful + freeze)
/pair-agent	Collab	Pair a remote AI with your browser
/skill-creator	Meta	Create and test new skills
/gstack-upgrade	Maintenance	Upgrade gstack to latest
/devex-review	DX Audit	Live developer experience testing

Creator Marketing (31)

Command	What It Does
/audience-demographic-analyzer	Analyze influencer audience fit
/brand-safety-screen	Flag controversial creator content
/campaign-brief-generator	Generate ready-to-send campaign briefs
/campaign-goal-to-kpi-framework-builder	Map objectives to KPIs
/campaign-roi-calculator	Calculate ROI with leadership narrative
/campaign-status-dashboard-digest	Weekly status for Slack/email
/content-approval-feedback-formatter	Turn internal notes into creator feedback
/content-capture-checklist-builder	Monitoring checklist per platform
/content-to-brief-compliance-checker	Pass/fail against brief requirements
/creator-briefing-faq-generator	Anticipate creator questions
/creator-content-concept-generator	5-10 native content ideas
/creator-negotiation-assistant	Rate negotiation with trade-off analysis
/creator-outreach-sequence-generator	Multi-touch outreach across platforms
/creator-posting-compliance-tracker	Track required vs actual posts
/creator-rate-estimator	Fair market rate estimation
/engagement-rate-calculator-benchmarker	Benchmark against platform averages
/ftc-disclosure-spot-checker	FTC compliance review
/metrics-normalization-formatter	Clean messy metrics into one table
/multi-platform-format-adapter	Adapt briefs for TikTok/IG/YT Shorts
/niche-fit-scorer	Score creator-brand alignment 1-10
/organic-repost-caption-writer	Repost captions with proper credit
/paid-ad-copy-adapter	Turn organic content into ad copy
/paid-social-creative-brief	Paid social brief from creator posts
/performance-benchmark-setter	Pre-launch KPI benchmarks
/post-campaign-creator-scorecard	Score creators after campaign
/quarterly-creator-program-review	QBR with wins, misses, pivots
/reply-triage-classifier	Classify influencer replies into actions
/story-metrics-screenshot-parser	Parse Story insights screenshots
/universal-creator-follow-up-chaser	Follow-up sequences for any scenario
/utm-parameter-builder	Generate UTM tracking URLs
/verbal-agreement-summarizer	Summarize DM/email deal terms

Marketing Suite (15)

Command	What It Does
/market	Main orchestrator — launches all skills
/market-ads	Ad creative and copy generation
/market-audit	Full marketing audit (5 parallel agents)
/market-brand	Brand voice analysis and guidelines
/market-competitors	Competitive intelligence analysis
/market-copy	Copywriting analysis and generation
/market-emails	Email sequence generation
/market-funnel	Sales funnel analysis and optimization
/market-landing	Landing page CRO analysis
/market-launch	Product launch playbook
/market-proposal	Client proposal generator
/market-report	Marketing report (Markdown)
/market-report-pdf	Marketing report (PDF)
/market-seo	SEO content audit
/market-social	Social media calendar and content

Design System (12)

Command	What It Does
/design	Comprehensive design skill (brand, tokens, logos, banners, slides)
/design-system	Token architecture + component specs
/frontend-design	Production-grade UI with high design quality
/canvas-design	Visual art in PNG/PDF
/banner-design	Social media, ads, web, print banners
/slides	Strategic HTML presentations with Chart.js
/ui-styling	shadcn/ui + Tailwind styling
/ui-ux-pro-max	50+ styles, 161 palettes, 57 font pairings
/web-design-guidelines	Web Interface Guidelines compliance
/brand	Brand voice and visual identity
/brand-context	Build brand context for all skills
/brand-guidelines	Apply brand colors and typography

GitNexus — Code Intelligence (7)

Command	What It Does
/gitnexus-cli	Index repos, check status, generate wiki
/gitnexus-exploring	Understand architecture and trace flows
/gitnexus-debugging	Trace bugs through the knowledge graph
/gitnexus-impact-analysis	Blast radius before editing code
/gitnexus-refactoring	Safe rename, extract, split, move
/gitnexus-pr-review	Review PRs with graph-aware context
/gitnexus-guide	Tools reference and schema docs

Best Practices (4)

Command	What It Does
/next-best-practices	Next.js file conventions, RSC, data patterns
/vercel-react-best-practices	React/Next.js performance from Vercel Engineering
/playwright-best-practices	Test writing, POM, CI/CD, flaky test fixes
/remotion-best-practices	Video creation in React

Utilities (6)

Command	What It Does
/pdf	Read, merge, split, watermark, create PDFs
/mcp-builder	Build MCP servers for external APIs
/doc-coauthoring	Structured doc/spec co-authoring workflow
/agent-reach	17 platforms via CLI, MCP, curl, Python
/superpowers	Process discipline (design-first, TDD, review)
/deploy-to-vercel	Deploy apps to Vercel

Speed Comparison

Task	Human Team	You + This System
Scaffold a feature	2 days	15 min
Write tests	1 day	15 min
Build end-to-end	1 week	30 min
Bug fix + regression test	4 hours	15 min
Full marketing audit	2 weeks	20 min
Creator campaign brief	3 hours	5 min

BRAIN Obsidian Vault

The BRAIN/ folder is an Obsidian vault with linked navigation:

BRAIN/
├── core/           — Master prompt, sprint workflow, control panel
├── navigation/     — Index, dashboard
├── systems/        — Architecture, AI Software Factory reference
├── tools/          — Tool nodes (gstack, marketing, etc.)
├── projects/       — Project template nodes
└── automations/    — Webhook maps, API plans

Open it in Obsidian: File > Open Vault > select the BRAIN folder.

Customization

After install, personalize your system:

1. Edit BRAIN/core/MASTER_PROMPT.md with your business context
2. Edit BRAIN/core/MASTER_MEMORY.md with your stack and preferences
3. Edit ~/CLAUDE.md with your project-specific routing rules
4. Add your projects to BRAIN/navigation/dashboard.md

Hard Rules (Built Into The System)

1. No code without a design. Ever.
2. No production code without a failing test.
3. No merge without review. Both stages.
4. No "done" without verification.
5. No fixing without investigation first.

License

Free to use. The underlying tools are MIT licensed:

• gstack: https://github.com/garrytan/gstack
• Superpowers: https://github.com/nicholasgriffintn/superpowers