Topic Gap Analyzer

What This Skill Does

Given:

1. A topic research PDF (from the user's Google AI Studio app, covering semantic analysis, related subtopics, entity salience, related keywords, and search volume context)
2. A URL inventory of the user's website — either an XML sitemap URL or a Screaming Frog CSV/XLSX export

…this skill analyzes the semantic gap between what the site currently covers and what the topic report recommends, then outputs a structured content action plan.

---

Step-by-Step Instructions

Step 1 — Parse the Topic Report PDF

Extract the following from the PDF:

• Core topic and its semantic definition (what the topic actually means in the current search landscape)
• Primary audience segments (e.g., new users vs. scaling brands)
• Related subtopics with their volume signals (HIGH / MEDIUM) and trend direction (Rising / Stable)
• Related keywords / search volume context — the keyword clusters and variants listed
• Top topic entities with their salience scores and entity type (Organization, Concept, etc.)
• Key concepts that modern content must address (e.g., "Lump Sum Trap", "payout-centric automation", "AI transaction categorization")

Entity handling rule: Entities that are Organizations (brands/competitors) must be flagged. Only include competitor/brand names in content recommendations if the page type explicitly calls for it — comparison pages, alternative pages, review roundups, "best X" lists. For all other page types, recommend covering the concepts those entities represent without naming the brands.

---

Step 2 — Ingest the URL Inventory

If the user provides an XML sitemap URL:

• Fetch the sitemap using the web_fetch tool
• If it's a sitemap index, fetch the child sitemaps and collect all <loc> URLs
• Extract all URLs into a flat list

If the user provides a Screaming Frog export (CSV or uploaded file):

• Read the file from /mnt/user-data/uploads/
• Extract the Address column (or equivalent URL column)
• Filter to 200 status codes only (ignore redirects, errors) if a Status column is present

Parse each URL to extract:

• The URL slug / path
• Any inferred page title or topic from the slug (split hyphens/underscores into readable words)

Do not fetch individual page content unless the user explicitly asks for deeper analysis — slug-level semantics are sufficient for gap analysis.

---

Step 3 — Semantic Matching

For each subtopic and keyword cluster from the report, scan the URL inventory for semantic matches using slug analysis:

• Strong match: The URL slug contains 2+ key terms from the subtopic/keyword cluster
• Weak match: The URL slug contains 1 key term that overlaps
• No match: No URL covers this subtopic

Also flag URLs that are adjacent — they cover a related theme and could be expanded rather than replaced.

Consider page type signals in URL slugs (e.g., /blog/, /vs/, /alternatives/, /reviews/, /best-) when making recommendations, as these affect whether competitor names are appropriate.

---

Step 4 — Generate the Content Action Plan

Structure the output as follows:

---

📊 Topic Overview

Brief 2–3 sentence summary of the topic, its current search intent landscape, and the two primary audience segments from the report.

---

✅ UPDATE Recommendations

Pages that already exist and should be enriched to cover the topic or subtopics better.

For each recommendation:

• URL: [existing URL]
• Why update: What the page currently covers and the gap
• What to add: Specific subtopics, keyword angles, or concepts to incorporate
• Entity requirements: Which non-brand entities/concepts from the report must appear
• Priority: High / Medium (based on subtopic volume + trend)

---

🆕 CREATE Recommendations

Net-new pages to build. Group by content type where relevant (e.g., educational guide, comparison page, feature landing page).

For each recommendation:

• Suggested URL: Proposed slug
• Page type: Educational / Landing / Comparison / Alternative / Hub etc.
• Target topic/subtopic: Which part of the report this addresses
• Target audience segment: Which persona this serves
• Core angle: The specific hook or framing for the page
• Must-cover entities/concepts: Non-negotiable topic entities from the report with high salience
• Competitor mentions: Only include if page type is Comparison / Alternative / Review / Best-list — flag clearly with ⚠️ Competitor mentions appropriate
• Priority: High / Medium / Low

---

🔑 Entity Coverage Checklist

A table of the top entities from the report and which recommended pages (UPDATE or CREATE) will cover them. This ensures no high-salience entity is left unaddressed.

Entity	Salience	Type	Covered By
...	...	...	...

---

📋 Quick Summary

• Total URLs analyzed: X
• Strong matches found: X
• Weak/adjacent matches: X
• UPDATE recommendations: X
• CREATE recommendations: X

---

Important Rules

1. Never suggest adding competitor brand names to non-comparative content. Instead recommend covering the function or concept they represent (e.g., "AI-powered transaction categorization" rather than naming a specific tool).

2. Prioritize Rising-trend subtopics over Stable ones when there are more opportunities than the site can realistically tackle at once.

3. Respect page type context: A /blog/ URL is a different recommendation context than a product landing page slug.

4. Don't hallucinate page content — all analysis must be based on what the slug implies, not assumed body content.

5. Be specific about what to add — vague recommendations like "cover this topic better" are not useful. Name the specific concepts, angles, and entity terms that should appear.

6. Hub vs. spoke thinking — if multiple subtopics cluster together with no existing hub page, recommend a pillar/hub page as a CREATE alongside the spoke pages.