Skill 詳細
website-cloner
End-to-end workflow to rebuild and improve a website from a URL.
使用前に確認
自動レビューは関連性のみを確認し、安全性や推奨を保証しません。使用前に出典の説明を読んでください。
SKILL.md
これはレビュー時に保存された抜粋です。完全で最新の内容は外部ソースを確認してください。
--- name: website-cloner description: "Build an improved website clone from a URL via 6-phase gated workflow (Vite/React/shadcn/Tailwind + GitHub Pages). Use for end-to-end site rebuilds. Don't use for single-phase work." license: MIT effort: high metadata: version: 1.2.1 author: "Luong NGUYEN <[email protected]>" --- # Website Cloner 6-phase orchestrator that clones any website and produces an improved version — better performance, UI/UX, SEO, and security — built with Vite + React + shadcn/ui + Tailwind CSS, deployable to GitHub Pages. ## When to Use Trigger when the user asks to: - Clone, rebuild, or recreate a website ("clone this site", "make a better version of <url>") - Improve a website's performance, UI/UX, or SEO by analyzing and rebuilding it - Start a full website improvement workflow from a URL Do **not** use for single-phase work (use the specific sibling skill directly). ## Prerequisites - A target URL (publicly reachable preferred). - Write access to create a local project dir (defaults under `~/workspace/clones` or `$CLONE_DIR`). - The sibling phase skills installed (website-analyzer, website-clone-report, etc.). - Optional: GitHub token if you want Pages deploy automation in Phase 5. - User approval at gates (explicit confirmation before Phases 3, 4, 5 advance). If a prerequisite is missing, stop and report it — do not guess paths or credentials. ## Workflow ``` Phase 1 — Analyze → website-analyzer Phase 2 — Report (gate) → website-clone-report Phase 3 — Propose (gate) → website-improvement-prd (outputs prd.md) Phase 4 — Plan (gate) → website-implementation-plan (outputs tasks.md) Phase 5 — Build → website-builder Phase 6 — Final Report → website-clone-final-report ``` Approval gates after Phase 2, 3, and 4: the orchestrator **must not advance** without explicit user approval. **Artifacts** (each written once, then referenced by name in the phases below): `analysis.json` — Phase 1's structured findings; `report.md` — Phase 2's plain-language summary; `prd.md` — Phase 3's improvement proposal; `tasks.md` — Phase 4's phased implementation plan, including the approved GitHub Actions artifact-deployment task; `builder-metadata.json` — Phase 5's build metadata, workflow-produced Pages URL, and structured post-deployment performance/SEO/security snapshot consumed by Phase 6; `after-analysis.json` — the comparable Phase 5 re-audit source. Phase 5 also produces base-aware Vite configuration and `.github/workflows/deploy-pages.yml`, which builds and deploys `dist/` rather than publishing the repository root. ## Layout This umbrella and its phase skills live together in a single suite folder: ``` skills/website-cloner/ ← this umbrella ├── SKILL.md ← orchestrator (you are here) ├── website-analyzer/ ← Phase 1 ├── website-clone-report/ ← Phase 2 ├── website-improvement-prd/ ← Phase 3 ├── website-implementation-plan/ ← Phase 4 ├── website-builder/ ← Phase 5 └── website-clone-final-report/ ← Phase 6 ``` Why nested: the phases are tightly coupled to this umbrella's data flow (analysis JSON → report → PRD → tasks → built site → final report). Keeping them in one folder makes the suite easy to browse, audit, and ship together. Each phase skill stays independently installable — the installers (`install.sh`, `remote-install.sh`) discover both top-level and nested skills. When invoking phase skills below, refer to them by name (`/website-analyzer`, `/website-clone-report`, …); the runtime resolves names regardless of filesystem path. See the individual phase skill docs for their full references/ and scripts/. This orchestrator stays short to fit the agent's context budget. ## Repo Sync Before Edits (mandatory) Before modifying files in a repository: ```bash branch="$(git rev-parse --abbrev-ref HEAD)" git fetch origin git pull --rebase origin "$GitHub で全文を読む (外部ページ)