- Home
- Skills
- Marketing & SEO
- SEO + AI-Search Audit: Scored Report and Fix List
More screenshots
Works with the AI Agents you already use
SEO + AI-Search Audit: Scored Report and Fix List
Ask any agent to "audit my page" and you get a good list of tips.
$5.99
SEO + AI-Search Audit: Scored Report and Fix List
Example session with this skill installed
Audit the sourdough-class page of a bakery (source and robots.txt pasted). I want it to win "sourdough class near me" and be cited by AI search.
- Read your context and instructions
- Compiled the seo + ai-search
SEO + AI-Search audit— sourdough-class page (fictional bakery)
Date: <audit date> · Pages audited: 1 (pasted source) · Goal: bookings (Inferred from "Book now") · Audience: local, English (country Unknown)
Target query: "sourdough class near me"
Executive summary
Provisional ceiling
30/100 (below 50: needs major work). The page has a generic title, two <h1> tags and no answer to what the class is, when, where or for how much. The pasted robots file blocks one AI crawler; intent is unknown.
Top 3 fixes (by impact ÷ effort): 1) real title and meta description 2) one <h1> and a question-style outline 3) decide the AI-crawler policy on purpose. Biggest lever by points: finding 5 (answer-first content), ranked lower only because it takes more work.
Not measured: status and redirects, sitemap, rendering, speed, rankings, backlinks, AI-citation rate. The paste may be partial (no head beyond <title> was shown).
Scorecard
Method: full points minus observed losses; Unknown items are not deducted, so the total is a provisional ceiling. The user said they pasted the page source, so head items missing from the paste (title, meta description, canonical, viewport, structured data) are treated as absent; the paste may be partial, and the report says so.
| Category | Score | Points lost and why (sub-weights from the rubric) |
|---|---|---|
| Crawl and index access | 12/15 (partly measured) | canonical 0/3 (none in the paste). Status, redirects, sitemap: Unknown (paste only). The GPTBot rule is not deducted anywhere: intent is Unknown (see finding 3) |
| Titles, meta, headings | 1/15 | title 1/5 (generic "Home") · meta description 0/4 (none in the paste) · single h1 0/3 (two) · outline 0/3 (none) |
| Content match to intent | 3/20 | answers the query 0/7 · sub-questions 0/5 · original information 0/5 · next step 3/3 ("Book now") |
| Trust and expertise | 1/10 | author/organization 0/3 · contact 0/2 · dates 0/2 · sources and no unsupported claims 1/3 ("best bakery in town!!!" is unsupported) |
| Structured data | 0/10 | none found |
| AI-search readiness | 4/15 | access 4/4 (one AI crawler disallowed: intent Unknown, not deducted; see finding 3) · answer-first 0/4 · entities 0/3 · trust/sources 0/2 · freshness 0/2 |
| Experience basics | 5/10 (partly measured) | viewport 0/2 (none in the paste) · images alt and size 0/3. Readability, overlays, performance: Unknown |
| Links | 4/5 (partly measured) | credible outbound 0/1 (none). Anchor text and broken-link checks: Unknown, the paste has no links |
Total: 30/100, a provisional ceiling (12+1+3+1+0+4+5+4). A total of 100 is reported only when every item was measured. The cap rule is not triggered because category 1 is 12/15.
Findings (ordered by impact ÷ effort; nothing in the paste stops crawling)
1. Generic title and no meta description
[Observed in the pasted source] <title>Home</title>; no meta description in the paste — head
Why it matters: the title is the page's label in results and one of the first things engines read.
Fix: replace with a descriptive title using the real class details, for example Sourdough Baking Class in [City] | [Bakery name] (placeholders; length <n> chars + placeholder text, keep the total at 60 or less).
Verify: view-source after publishing; Search Console URL Inspection.
Impact 3 · Effort 1
2. Two <h1> and no outline
[Observed] two <h1> elements, no <h2> — body
Why it matters: one clear topic heading and question-style subheadings are how both readers and engines scan the page.
Fix: one <h1> "Sourdough Baking Class in [City]"; <h2> for "What you will learn", "Dates and price", "Where we are", "Questions".
Verify: view-source heading outline.
Impact 2 · Effort 1
3. AI-crawler policy needs a decision
[Observed in the pasted robots.txt] GPTBot is disallowed for the whole site; the paste may be incomplete
Why it matters: this keeps that crawler away from every page. GPTBot is generally described as a training crawler rather than a search crawler; confirm in the vendor's documentation. Some owners want that, some want visibility in AI answers.
Fix: ask the owner. If visibility is wanted, remove the rule after confirming current crawler names and purposes in each vendor's documentation.
Verify: fetch /robots.txt yourself after the change; server logs.
Impact 2 · Effort 1
4. Unsupported superlative
[Observed] "We are the best bakery in town!!!" — body
Why it matters: it is a claim with no source and reads as filler to readers and quality systems.
Fix: replace with a checkable fact: years baking, number of classes taught (only if true), a dated customer quote with permission. I will not write such facts for you.
Verify: keep the evidence for every claim that remains.
Impact 2 · Effort 1
5. No answer-first content
[Observed] the body is one slogan sentence — body
Why it matters: nothing here can be quoted as an answer to "sourdough class near me".
Fix: under the <h1> add 2-3 sentences with the real details: what the class teaches, [duration], [dates], [address], [price], how to book. Add a small table of upcoming dates once they exist. Use only facts the owner confirms.
Verify: read the first screen on a phone; does it answer the query? Re-run this audit.
Impact 3 · Effort 2
6. No structured data
[Observed] none in the paste — head
Why it matters: structured data helps engines tie the page to a business and an event.
Fix draft (only after the details are visible on the page): LocalBusiness for the bakery and Event for each class date, using the visible address, dates and price. Do not add ratings unless real reviews are shown.
Verify: a schema validator and the search engine's own testing tool.
Impact 2 · Effort 2
7. No canonical and no viewport meta
[Observed in the pasted source] neither <link rel="canonical"> nor <meta name="viewport"> appears — head
Why it matters: a canonical tells engines which URL is the main one; the viewport tag controls how the page renders on phones.
Fix: add <link rel="canonical" href="[final page URL]"> and <meta name="viewport" content="width=device-width, initial-scale=1">.
Verify: view-source; a mobile preview.
Impact 1 · Effort 1
Drafts for your review (not applied)
- Title:
Sourdough Baking Class in [City] | [Bakery name]—<n> chars + placeholder text - Meta description: "Learn sourdough at [Bakery name] in [City]. [Level], [duration], [price]. See dates and book." — fill the placeholders from real data, then count (target 70-160)
- Heading outline: H1 Sourdough Baking Class in [City] · H2 What you will learn · H2 Dates and price · H2 Where we are · H2 Questions
Connects securely to your tools. The creator never sees your data.
About this skill
Ask any agent to "audit my page" and you get a good list of tips. This skill makes the audit a repeatable, client-ready deliverable:
- A 100-point rubric with sub-weights (crawl access, titles/meta/headings, content vs intent, trust, structured data, AI-search readiness, experience basics, links). Every point lost has a reason.
- An AI-search checklist: crawler access, answer-first passages, entity clarity, trust signals, freshness, and how to measure honestly.
- Every finding says what was seen, why it matters, the exact fix and how to verify it, and is labelled Observed / Inferred / Unknown, so a guess is never presented as a fact.
- Prioritized by impact vs effort, blockers first. Drafts (title, meta, outline, JSON-LD) are offered, never applied without your OK.
- A report template and a complete worked example.
What you get: SKILL.md (procedure), a scoring rubric with sub-weights, an AI-search checklist, a structured-data cheat sheet, a report template and a worked example.
Honest limits: no rankings, traffic or citation numbers are invented; nobody can guarantee rankings or AI citations; it audits pages you name (max 10) and does not crawl at scale; it cannot run validators or speed tests for you, but tells you which to run; length figures are common heuristics, not search-engine rules.
Privacy: the skill installs nothing and makes no network calls itself.
Maximum 10 pages per run. No access to private analytics, traffic volume, or keyword difficulty data. Cannot perform bulk crawling or bypass authentication.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed today
Permissions
The skill is instructions and reference files only and contains no code. Network Access: if you ask your agent to audit a URL, the agent's own web tools fetch only the pages you name (max 10); otherwise you paste the page source. Write Files: by default nothing is written; the agent edits your site files only if you ask it to apply a draft, and it shows the exact change first.