<!-- GENERATED from AGENTS.md by scripts/sync-skills.mjs — edit the source, then run: node scripts/sync-skills.mjs -->

ai-site-cloner

Template for cloning any multi-page website into this Next.js codebase, then rebranding it as your own. Two commands drive everything: /clone-website <url> and /restyle.

Tech Stack

• Next.js 16 — App Router, React 19, TypeScript strict
• Tailwind CSS v4 + shadcn/ui (base-ui primitives), oklch tokens in src/app/globals.css
• Lucide React icons (replaced by extracted SVGs in src/components/icons.tsx during cloning)
• Playwright (devDependency) powers all extraction/QA scripts

Commands

• npm run dev / npm run build / npm run lint / npm run typecheck
• npm run check — lint + typecheck + build (must pass before any commit)

Architecture Rules

• Content lives in src/data/*.ts, never hardcoded in components. Components take props. This is what makes /restyle possible — copy swaps without touching layout.
• Pages: src/app/<route>/page.tsx wires data → components. Shared header/footer mount in layouts.
• docs/research/ holds extraction artifacts (JSON, specs, manifest) — auditable, never hand-edited except spec files.
• docs/research/manifest.json is the pipeline state. Update it at every stage transition; check it first on any clone-related task (node scripts/manifest.mjs status).
• Specs (docs/research/components/**/*.spec.md) must pass node scripts/lint-spec.mjs before any builder agent is dispatched.
• All extraction numbers come from scripts/ (computed styles, real column counts, pixel diffs) — never estimated by eye.

Code Style

• TypeScript strict, no any
• Named exports, PascalCase components, camelCase utils
• Tailwind utility classes; arbitrary values (text-[17px]) when the extracted value doesn't match Tailwind's scale
• 2-space indentation

Skills / Commands

Two workflows, each with a single source of truth in .claude/skills/<name>/SKILL.md:

| Command | Purpose | |---|---| | /clone-website <url> | Crawl, extract, spec, build, and pixel-diff QA a full multi-page clone | | /restyle | Swap colors, fonts, logo, and copy for your brand — layout stays intact |

Where your tool finds them (all generated from the same source):

• Claude Code: .claude/skills/<name>/SKILL.md
• Codex CLI: .codex/skills/<name>/SKILL.md
• GitHub Copilot coding agent: .github/skills/<name>/SKILL.md
• Cursor: .cursor/commands/<name>.md
• Windsurf: .windsurf/workflows/<name>.md
• Gemini CLI: .gemini/commands/<name>.toml
• OpenCode: .opencode/commands/<name>.md
• Continue: .continue/commands/<name>.md
• Augment Code: .augment/commands/<name>.md
• Amazon Q Developer CLI: .amazonq/cli-agents/<name>.json
• Cline / Roo Code, Aider, anything else: open .claude/skills/<name>/SKILL.md and follow it as a prompt.

The extraction scripts are plain Node CLIs any agent can run; Chromium auto-installs on first use (or npm run setup). No subagent support in your tool? Build sections sequentially from the same lint-passing specs.

Ethics

Clone sites you own or have permission to rebuild (platform migration, lost source, learning). Never for phishing, impersonation, or passing off someone's design as your own.

Keeping Agent Configs in Sync

• .claude/skills/*/SKILL.md and this file (AGENTS.md) are the only sources of truth for per-tool configs.
• After editing either, run node scripts/sync-skills.mjs to regenerate every per-tool copy.
• Generated files are committed; CI runs node scripts/sync-skills.mjs --check and fails if any are stale.