- Home
- Skills
- Marketing & SEO
- AI Search Readiness Check + llms.txt Generator
AI Search Readiness Check + llms.txt Generator
Every token was checked against the vendor's own documentation on 23 Oct 2026.
$2
Works with the AI Agents you already use
AI Search Readiness Check + llms.txt Generator
Example session with this skill installed
Check https://plausible.io for AI-search readiness and write me an llms.txt and the JSON-LD I'm missing. Then quickly check our demo site https://remont.144-31-16-235.sslip.io too.
- Read your context and instructions
- Compiled the ai search readiness
Real run on 23 Oct 2026: scripts/ai_readiness_check.py on both sites (homepage + 5 key pages for Plausible; the demo site is one page), judged with references/checks.md and written up with the report template. Shortened for the listing. Your results will differ as sites change.
AI-search readiness: plausible.io — 85/100 · Ready
Checked 23 Oct 2026 · homepage + 5 key pages (about, contact, features, imprint, docs)
| # | Check | Status | Finding |
|---|---|---|---|
| A1 | Reachable | PASS | HTTPS, 200 in 0.4 s, no challenge |
| A2 | AI search crawlers in robots.txt | PASS | robots.txt has only a Sitemap: line, so all crawlers are allowed |
| A3 | Indexable / snippet-able | PASS | no noindex / nosnippet in meta or headers on 6 pages |
| A4 | Sitemap | PASS | declared in robots.txt, 166 URLs with <lastmod> |
| B1 | Title | WARN | homepage title is 75 characters, so it gets cut off |
| B2 | Meta description | PASS | 151 characters, specific |
| B3 | H1 & headings | WARN | one H1, but 3 menu <h3>s come before it; /imprint has no H1 |
| B4 | Server-rendered content | PASS | ~1,000 words in raw <main>; plans and prices are in the HTML |
| B5 | lang & canonical | PASS | lang="en", self-canonical on every page checked |
| B6 | Key facts in text | PASS | what ✔ who ✔ where ✔ price ✔ contact ✔ |
| B7 | JSON-LD | WARN | only WebSite / WebPage / BreadcrumbList, and BlogPosting on /features; no Organization, no offers for the visible plans |
| C1 | /llms.txt | FAIL | /llms.txt and /docs/llms.txt return 404 |
Fix first: 1) publish the llms.txt below; 2) add Organization + SoftwareApplication JSON-LD and change BlogPosting on /features to WebPage; 3) shorten the title and turn the menu <h3> labels into plain text.
For your information: training crawlers (GPTBot, ClaudeBot, Google-Extended, Applebot-Extended) are allowed, which is your choice. CDN bot rules can't be seen from outside. No one can guarantee AI citations; these fixes remove obstacles.
Generated /llms.txt (excerpt; every fact is from plausible.io pages fetched that day):
# Plausible Analytics
> Lightweight, open-source, privacy-friendly alternative to Google Analytics: no cookies, no personal data collected, made and hosted in the EU.
- Who: Plausible Insights OÜ, Tartu, Estonia: an independent, self-funded company, team of 10, started in December 2018.
- Open source: AGPLv3, code at https://github.com/plausible/analytics
- Pricing (per month, up to 10k monthly pageviews, as shown on the homepage): Starter $9, Growth $14, Business $19; Enterprise custom. 30-day free trial.
- Contact: hello@plausible.io
## Product
- [Features](https://plausible.io/features): AI traffic analytics, segments, real-time, funnels, goals, revenue tracking
- [Pricing](https://plausible.io/#pricing): plan comparison
## Docs
- [Documentation](https://plausible.io/docs): setup guides, dashboard, settings
## Company
- [About](https://plausible.io/about) · [Contact](https://plausible.io/contact) · [Imprint](https://plausible.io/imprint)
## Optional
- [Blog](https://plausible.io/blog): guides and product updates
AI-search readiness: our demo repair-shop site — 40/100 · Blocked: invisible to AI search
robots.txt has User-agent: * + Disallow: /, and the homepage has <meta name="robots" content="noindex"> — correct while it is a demo, remove both at launch. Passes: one clear H1, ~950 words of server-rendered Russian text with full price tables. Fails: no JSON-LD (LocalBusiness suggested), no llms.txt. The generated Russian llms.txt uses the real services, hours, prices and warranty, and writes [TODO: адрес — на сайте указан пример] instead of copying demo placeholders as facts.
Connects securely to your tools. The creator never sees your data.
About this skill
Give your agent a website URL and ask: "Can AI assistants find and cite this site?" It reads the homepage and a few key pages (about, contact, pricing, services, legal, FAQ, docs), then returns a short scored report and the files to fix what it found.
12 checks, each PASS / WARN / FAIL with the exact fix:
- robots.txt for AI crawlers — verdicts per token under RFC 9309 matching rules, for the search and answer crawlers (OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot, Bingbot, Applebot) and the user-initiated fetchers (ChatGPT-User, Perplexity-User, Claude-User). Every token was checked against the vendor's own documentation on 23 Oct 2026.
- Training opt-outs reported separately — GPTBot, ClaudeBot, Google-Extended and Applebot-Extended are shown as info, because blocking model training is your choice and does not remove you from AI search.
- Indexability —
noindex,nosnippetandmax-snippetin meta tags andX-Robots-Tagheaders on every page checked. - Sitemap, title, meta description, one clear H1, heading order,
lang, canonical. - Server-rendered vs JS-only content — word count of the raw HTML, the way many AI crawlers see it, not a browser-rendered copy.
- Plain-text facts — is it clear what you do, who you are, where you work, what it costs and how to contact you, in text a machine can quote?
- JSON-LD — Organization / LocalBusiness / Product / SoftwareApplication / FAQPage present, valid, and matching the visible page.
- /llms.txt — present and valid under the llmstxt.org proposal (also checks
/docs/llms.txt).
Generated files:
- a ready
llms.txtin llmstxt.org format: H1, summary, key facts, link sections,## Optional; - an optional
llms-full.txtoutline; - suggested JSON-LD blocks for your pages.
No invented facts. Every fact in the generated files comes from a page the agent fetched, with its source URL. Anything missing becomes a visible [TODO: …] placeholder. If your site marks something as an example (a demo address, "legal details will be here"), the agent leaves it out and flags it.
Honest by design. No one can guarantee that an AI assistant will cite your site, and the skill says so in every report. It never claims ranking effects it can't back up. It also tells you that Google says no special files or schema are needed for AI Overviews, so llms.txt is a cheap, low-risk extra for tools that read it, not a magic switch. The score weights are the skill's own editorial judgement, and it labels them that way.
Read-only and polite. About 10–15 requests per audit. No logins, no form submissions. If a site shows a bot challenge, the agent reports it and does not work around it.
Works in any agent that can fetch web pages: Claude Code, Cursor, Codex, Windsurf, VS Code Copilot and others. The bundled Python helper (standard library only) collects the raw facts in one command. Without Python, the agent uses curl or its own web-fetch tool.
- It samples the homepage and up to about 6 key pages. It does not crawl the whole site.
- It does not measure actual citations, rankings or AI traffic.
- CDN/WAF bot blocking, IP-based rules and server logs are invisible from outside. The report reminds you to check them.
- Sites behind a bot challenge or login can't be audited. The agent reports the block and stops.
- JSON-LD suggestions aren't validated against Google's rich-result rules. Use Google's Rich Results Test or validator.schema.org.
- The crawler list was verified on 23 Oct 2026. Vendors add and rename bots, and the skill only lists tokens confirmed in vendor documentation. Microsoft Copilot-specific controls beyond Bingbot were not confirmed.
- The score weights are the skill's own judgement, not a published ranking formula.
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 25 days ago
Requires
Permissions
Allowed Hosts
Network: read-only HTTPS GET requests to the website being audited, about 10–15 per run (homepage, up to 6 key pages, robots.txt, sitemap, llms.txt, llms-full.txt, docs/llms.txt). There are no other hosts, no telemetry and no API keys. The helper identifies itself with an honest user agent ("ai-search-readiness-check … site owner self-audit"), does not impersonate AI crawlers and does not bypass bot protection. It writes a JSON file only if you pass --out. It never logs in or submits forms, and the agent does not publish or upload the generated files. The package is text only: Markdown and one Python script, no secrets.