# AI visibility checker for ChatGPT, Claude and Perplexity

> Check whether AI search tools and assistants can reach your page, which bots your robots.txt blocks, and what text they actually read when they do.

Check a URL: https://sharelint.com/?view=ai&url=<your-url>, or as Markdown: https://sharelint.com/api/preview?format=md&url=<your-url>

## Bots checked

- `OAI-SearchBot` (OpenAI): ChatGPT search. Search
- `ChatGPT-User` (OpenAI): ChatGPT browsing. On request
- `GPTBot` (OpenAI): Model training. Training
- `Claude-SearchBot` (Anthropic): Claude search. Search
- `Claude-User` (Anthropic): Claude web fetch. On request
- `ClaudeBot` (Anthropic): Model training. Training
- `PerplexityBot` (Perplexity): Perplexity search. Search
- `Perplexity-User` (Perplexity): Perplexity live fetch. On request
- `Google-Extended` (Google): Gemini training and grounding. Training
- `Applebot-Extended` (Apple): Apple Intelligence training. Training
- `meta-externalagent` (Meta): Meta AI training. Training
- `CCBot` (Common Crawl): Open web crawl used by many models. Training

## What the check covers

- robots.txt rules for each AI crawler, with the exact rule that allows or blocks it.
- A live request with each bot’s User-Agent, to catch servers and CDNs that block AI bots.
- noindex and other robots directives in meta tags and the X-Robots-Tag header.
- The text in your raw HTML. AI crawlers do not run JavaScript, so a page rendered in the browser can look empty to them.
- Headings, meta description and JSON-LD structured data.
- Whether /llms.txt exists and follows the proposed format.

## Search, on-request and training bots

AI companies run separate bots for separate jobs. Search crawlers index pages so an assistant can find and cite them. On-request fetchers read a page live when a person asks the assistant to. Training crawlers collect pages for future models.

Blocking a training crawler keeps your content out of model training, and does not remove you from AI answers. Blocking a search crawler does remove you from that assistant’s answers. The score counts search and on-request access and ignores training crawlers, which are a publisher’s choice.

## Common problems

- The page is built by JavaScript: AI crawlers see an empty shell. Render or pre-render the content on the server.
- A CDN bot setting blocks AI crawlers even though robots.txt allows them. On Cloudflare, check Security > Bots and AI Crawl Control.
- A broad "Disallow: /" in the User-agent: * group blocks every AI bot that has no group of its own.
- robots.txt returns a server error. Crawlers then treat the whole site as off limits.
- A noindex directive left over from staging.

## What it does not do

It does not predict what a model will say about your page. Answers change between runs and models, so they cannot be tested reliably. The check covers what is deterministic: whether each bot can reach the page, and what it reads.

Live requests come from Cloudflare’s network with each bot’s User-Agent. A site that verifies bots by IP address may treat the real bot differently.