AI Search Readiness Check
Find what to fix on a page before chasing AI visibility. Get evidence-linked findings, supported answer rewrites and practical verification steps, without invented scores or indexing promises.
$4
Works with the AI Agents you already use
AI Search Readiness Check
Example session with this skill installed
Original fictional raw user request
We run the public FieldLog help center. Please review this supplied HTML export and robots policy for the question: “How long can I download an activity export, and will exporting remove my records?” We want the help page available in public search but do not want to allow GPTBot for training. The noindex tag was accidentally copied from our staging template. Give us a small developer/editor repair queue, not a whole-site strategy. Do not edit the site.
These files are fictional fixtures prepared for the Skill demonstration. They were not fetched from a live website. The URLs use the reserved .example domain. Capture time, HTTP status and response headers are unknown. The “reviewed” date and product details are fictional page content, not independent verification.
page.html
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<title>Information | FieldLog Help</title>
<meta name="robots" content="noindex, follow">
<link rel="canonical" href="https://fieldlog.example/help/export-logs">
<script type="application/ld+json">
{"@context":"https://schema.org","@type":"FAQPage","mainEntity":[{"@type":"Question","name":"How long is the activity export link available?","acceptedAnswer":{"@type":"Answer","text":"Your export link is available for 30 days."}}]}
</script>
</head>
<body>
<nav><a href="/help">Help home</a></nav>
<main>
<h1>Activity exports</h1>
<p>FieldLog keeps your team's work organized. This article explains activity exports.</p>
<p>Reviewed by the FieldLog Support team on <time datetime="2026-10-30">30 October 2026</time>.</p>
<h2>Before you start</h2>
<p>Only workspace owners and admins can create an activity export. Each export contains one workspace's activity records.</p>
<h2>Download availability</h2>
<p>An activity export download link remains available for 48 hours after the export finishes. If the link expires, an owner or admin can create a new export.</p>
<h2>Your original records</h2>
<p>Creating or downloading an export does not delete the original activity records. Timestamps in the exported CSV use UTC.</p>
<p>Learn about <a href="/help/roles">workspace roles</a>.</p>
</main>
<footer><a href="/contact">Contact FieldLog Support</a></footer>
</body>
</html>
robots.txt
User-agent: *
Disallow: /private/
User-agent: OAI-SearchBot
Disallow: /help/
Allow: /help/export-logs$
User-agent: GPTBot
Disallow: /
Sitemap: https://fieldlog.example/sitemap.xml
input.json
{
"html": {
"file": "page.html",
"url": "https://fieldlog.example/help/export-logs",
"mode": "supplied_html",
"captured_at": null,
"tool": null,
"http_status": null,
"capture_status": "unknown",
"headers": []
},
"robots": {
"file": "robots.txt",
"url": "https://fieldlog.example/robots.txt",
"mode": "supplied_text",
"captured_at": null,
"tool": null,
"http_status": null,
"capture_status": "unknown",
"headers": []
}
}
- Read your context and instructions
- Compiled the ai search readiness
- Generated the document
FieldLog help-page repair pack
Demonstration based on original fictional supplied inputs, not a live audit.
Scope and evidence
Review question: “How long can I download an activity export, and will exporting remove my records?” The supplied request says this page should be public in search, the staging noindex is accidental, and GPTBot should remain blocked. No site edits were requested or made.
| ID | Evidence | Mode / date / status |
|---|---|---|
| R0 | request.md, user brief | Supplied fictional intent; no live site facts verified |
| H1 | page.html, 28 lines, claimed URL https://fieldlog.example/help/export-logs | Supplied HTML; capture time and HTTP status unknown; no response headers |
| R1 | robots.txt, 11 lines, claimed URL https://fieldlog.example/robots.txt | Supplied text; capture time and HTTP status unknown |
| E1 | evidence.json | Actual local helper output; contains input SHA-256 values and extraction limits |
The HTML contains readable main text, an identified support-team reviewer and a dated review. The biggest missing evidence is a current response capture with headers. These files cannot establish current accessibility or indexing.
Prioritized changes
| ID / priority | Observed evidence | Interpretation | Concrete change | Owner and acceptance test |
|---|---|---|---|---|
| F1 / P0 | H1 line 6: noindex, follow; R0 explicitly calls it an accidental staging directive | Conflicts with the requested public-search intent for crawlers honoring that meta directive. Does not prove the live page is currently excluded. | Remove noindex from the production template for this public page, for example <meta name="robots" content="follow">. Check that no applicable X-Robots-Tag header independently keeps noindex. Preserve private/staging exclusions. | Developer: inspect a fresh final-URL response and all applicable meta/header controls; confirm the public page has no unintended noindex. This is a directive test, not an indexing result. |
| F2 / P1 | H1 line 9, mainEntity[0].acceptedAnswer.text, says “30 days”; H1 line 21 says “48 hours after the export finishes” | Material contradiction inside otherwise parseable JSON. The helper confirms syntax only. The current headings are not an explicit visible FAQ matching that markup. | Remove this misleading FAQPage script. Retain the supported 48-hour explanation in the visible article. Reintroduce appropriate markup only if it accurately describes the final page and its purpose. No schema is needed to complete the immediate factual correction. | Developer/editor: confirm no remaining structured-data statement says 30 days; compare any retained markup against visible facts. Do not call this a rich-result validation. |
| F3 / P2 | H1 line 5: Information | FieldLog Help; H1 lines 15–23 contain the actual topic and details | The title is less descriptive than the page. The user's two-part answer is split between sections. | Use `<title>Activity export links and records | FieldLog Help</title>` and replace the generic opening paragraph with the answer below. Keep role restrictions in the nearby “Before you start” section. |
F1 is highest priority because public visibility intent is explicit. It is not permission to loosen unrelated policies. F2 is a correctness problem even if the page never appears in an AI answer.
Copyable opening answer
A FieldLog activity export link is available for 48 hours after the export finishes. Creating or downloading the export does not delete your original activity records. If the link expires, a workspace owner or admin can create a new export.
Fact map: sentence 1 and the expired-link remedy come from H1 line 21; non-deletion comes from H1 line 23; owner/admin scope is also supported by H1 line 19. This rewrite adds no performance, privacy, retention or indexing promise. It does not claim records can never be deleted by another action.
Keep and interpret carefully
Robots: R1 lines 4–6 name OAI-SearchBot and include both a /help/ block and a more-specific /help/export-logs$ allow rule. Under the documented matching model, the exact supplied path has an allow exception. Do
not diagnose this exact URL as blocked merely from line 5. The $ matters: a query-string or other URL variant needs a separate match check. This remains a conditional policy interpretation, not proof of a bot visit.
Training choice: R1 lines 8–9 block GPTBot. Keep that deliberate policy. Search and training settings are distinct.
Default group: R1 lines 1–2 disallow /private/, not this public help path. No Bing-specific rule is supplied. This cannot establish Bing crawling or indexing.
Readable text: H1 lines 19, 21 and 23 contain the key information directly in the supplied main HTML. No rendered-browser test was performed.
Accountability: H1 line 17 identifies the support-team reviewer and an explicit review date. The contact link is at line 26. These are present, not independently authenticated; there is no need to invent an expert biography.
Sources: the roles link at H1 line 24 is relevant to permissions, but its destination was not opened. The supplied article is the source of the fictional product facts, not external proof of how the real service behaves.
Canonical/sitemap: H1 line 7 identifies the intended canonical URL; R1 line 11 supplies a sitemap location. Neither proves an engine selected that canonical or ingested the sitemap.
Verify next
- Developer: capture the final live URL, response status, response headers and response HTML with time/tool details. Confirm it is the intended article, not a challenge or redirect landing page.
- Developer: capture the correct origin's current robots file and status; re-check the exact canonical path and any important query variants. Compare current policy with the supplied file before making changes.
- Editor/developer: verify the published content and removal or correction of misleading structured data. An authorized platform validator is a separate optional test, not already completed here.
- Owner: if actual indexing or citations matter, use property-authorized inspection and separately dated search observations. Record the engine, query, time and URL. Do not infer these outcomes from this review.
Outcome: three evidence-linked changes proposed; none applied. Live HTTP behavior, rendered DOM, Bing/Google indexing, ChatGPT citations, traffic and revenue were not tested. No numeric readiness score or monitoring promise is made.
page.html
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Your page may look fine in a browser while its exported HTML, crawler controls or structured data tell a different story. AI Search Readiness Check helps a compatible assistant separate what the evidence shows from what still needs testing.
Give it a URL with permitted browsing tools, or supply HTML and robots text for an offline review. It checks five areas:
- Crawler access and indexing/snippet controls, with search and training preferences kept separate.
- Readable main content, titles, headings and important information hidden behind scripts or interactions.
- Structured-data syntax and consistency with the actual page.
- Accountable authorship, meaningful dates and support for factual claims.
- Clear answer passages built only from your supported facts.
You receive a prioritized repair queue with exact evidence locators, concrete changes, owner roles and acceptance tests. When the source supports it, you also get a copyable answer passage. When evidence is missing, the report names the specific input needed instead of inventing a result.
Included
- The English Skill and decision guides.
- An optional Python evidence extractor with no external packages.
- Two original fictional worked examples, their raw inputs and generated evidence outputs.
- Executable helper tests and behavioral evaluation cases.
Requirements and limits
Use a compatible assistant that can read Skill files. Python 3.10+ is needed only for the optional local extractor. Live review requires separately available, permitted browsing tools; no paid API, crawler service or browser is included. The extractor itself is offline and does not execute JavaScript.
This is a focused page-readiness audit, not a backlink service, full technical crawl, legal certification or search analytics platform. It does not measure citation probability, promise indexing or rankings, validate every schema feature, change your site or publish anything automatically.
Example prompt
“Review this public help page and supplied robots file for our customer question. Keep our training opt-out. Show exact evidence, prioritize repairs, and give me one supported answer rewrite. Do not edit the website.”
No citation score, indexing/ranking guarantee, full-site crawl or automatic site edits. Offline HTML cannot prove live HTTP behaviour or rendered output. Extractor does not execute JavaScript or fully validate every schema feature. Live checks need separate permitted host tools.
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 8 days ago
Requires
Permissions
Offline mode reads user-supplied HTML/robots and writes local reports using optional Python. Browser/network only when the buyer requests a permitted live URL audit and the host provides access. No account credentials, website edits, publishing or paid API calls are required.