O — OBJECTIVE
客戶品牌的目標受眾在用 ChatGPT、Perplexity、Google 搜尋該領域專業問題時，客戶品牌網站是被引用和排名最前面的來源 — 不是因為 SEO 技巧，而是因為每篇文章的法規引用密度和專業深度讓 AI 引擎和搜尋引擎都認定它是最可信的權威。
G — GOAL
客戶品牌網站的 AI 爬蟲訪問量每週增加 20%，SEO 自然流量月增 15%，AI citation（被 ChatGPT/Perplexity 引用次數）每月 ≥5 次。
Team-specific angle: By rewriting existing website content into AI-friendly structured format, increasing the total volume of AI-citable content on the client brand's website.
### Strategy Rationale (why this team advances G)
透過改寫現有內容為 AI 友善格式，增加 AI 可引用的內容總量。
Operational target (from previous method-level G): 在 4 小時內（≤50 頁站點）將任何網站的內容從「只對瀏覽器友善的 HTML」轉變為「AI 引擎可直接引用的結構化知識」— 30 天內至少 1 篇鏡像文章被 AI 引擎引用，content-map.json 準確度 ≥95%。每篇鏡像文章具備原文連結、FAQ Schema、AEO 格式。
Success Criteria by Phase:
- Phase 1 (首站): ≤50 頁完成鏡像，4 小時內（見 Phase 1 Launch Checklist）
- Phase 2 (30 天後): ≥1 AI citation traceable to /mirror/ URL（自動化驗證：Perplexity API query 10 target keywords → parse source URLs → match against source-map.json → log to docs/citation-tracking/）
- Phase 3 (90 天後): ≥3 站點鏡像完成，月均 ≥5 AI citations
Phase 1 Launch Checklist (must complete before first publish):
``
Pre-publish:
[ ] Agent 1 scan-results.json: coverage ≥90% vs sitemap.xml count
[ ] Agent 2 extracted.md: spot-check 3 pages — noise <5%, all 5 metadata fields present
[ ] Agent 3 rewrite ratio: difflib check on all articles — all >60% before proceeding
[ ] Agent 4 schema.json: JSON-LD validator exit 0 on all articles
[ ] Agent 5: source-map.json complete — every /mirror/ URL has source_url, last_synced, rewrite_ratio, faq_count
Post-publish:
[ ] curl -s {mirror_url} | grep "Original source:" — source link banner present
[ ] curl -s sitemap.xml | grep "/mirror/" — at least 1 mirror URL indexed
[ ] curl -s /llms.txt | grep "/mirror/" — mirror section present
[ ] robots.txt allows GPTBot, ClaudeBot, PerplexityBot, Applebot (check all 4)
[ ] Perplexity manual check: search 1 target keyword — verify /mirror/ URL appears within 7 days
G Measurement Automation:python``
# scripts/check-citations.py — run monthly via cron
import json, requests
with open("source-map.json") as f:
mirrors = json.load(f)["mirrors"]
# Build keyword list from source-map categories
keywords = [m["source_url"].split("/")[-1].replace("-", " ") for m in mirrors[:10]]
citation_count = 0
for kw in keywords:
# Query Perplexity API (or scrape Perplexity search page)
# Check if any /mirror/ URL appears in citations
resp = requests.get(f"https://www.perplexity.ai/search?q={kw}", headers={"User-Agent": "Mozilla/5.0"})
if any(m["mirror_url"] in resp.text for m in mirrors):
citation_count += 1
# Log result
with open(f"docs/citation-tracking/{date.today()}.md", "w") as f:
f.write(f"# Citation Check {date.today()}\n- Keywords checked: {len(keywords)}\n- Citations found: {citation_count}\n")
# Phase 2 target: citation_count >= 1
# Phase 3 target: citation_count >= 5/month
Usage Boundary: 本系統主要用於鏡像自有品牌網站（如 clientbrand.com → clientbrand.ai）。若用於第三方網站，必須取得內容授權或確認 fair use，並在每篇文章加註來源授權聲明。
Agent 0: Team Leader
Input: Source website URL + target AI site path
Output: Pipeline orchestration decisions + final acceptance signal + citation tracking entry
Agent 1: Site Scanner（站點掃描器）
Agent 2: Content Extractor（內容萃取器）
Agent 3: Content Rewriter（內容改寫器）
Agent 4: Schema Builder（結構標記器）
Agent 5: Publisher（發布管理器）
source-map.json for all mirrored pageslast_synced timestampsource_removed, add notice banner to mirror pagesource-map.json with new last_synced timestampsource-map.json schema:
``
json
{
"mirrors": [
{
"source_url": "https://example.com/products/widget",
"mirror_url": "/mirror/products-widget/",
"category": "product",
"last_synced": "2026-04-22T08:00:00Z",
"status": "active",
"rewrite_ratio": 0.72,
"faq_count": 4
}
],
"last_full_scan": "2026-04-22T08:00:00Z",
"total_mirrored": 45,
"total_source_pages": 50
}
``