Check links and messages from your code or your AI agent. It's the same check as the website: free, up to 20 checks a day.
Just adding URLert to your own assistant? Add URLert to your AI assistant
MCP server
For AI agents. It signs in with OAuth.
API
For your code. It takes an API token.
Add the address to your MCP client as a remote server. It asks you to sign in to URLert (or make a free account), then works as you. It can't delete a check or your account.
The API has the same six operations, by the same names.
Reading our site instead? /llms.txt lists our pages. Add .md to any page's address for its Markdown.
Make a free account, then a token: Account, then AI agents, then New API token. It asks you to confirm it's you, and shows the token once.
Each operation is a POST to its name, with a JSON body. Send the token as a Bearer token.
curl https://www.urlert.com/api/v1/check_link \
-H "Authorization: Bearer $URLERT_TOKEN" \
-H "Content-Type: application/json" \
-d '{"link": "https://txtag-tolls.icu/pay"}'It answers at once, with the check's id and "status": "queued".
A check usually ends within a minute. Call get_result every few seconds, until status is done or failed.
curl https://www.urlert.com/api/v1/get_result \
-H "Authorization: Bearer $URLERT_TOKEN" \
-H "Content-Type: application/json" \
-d '{"id": "k3m9x2pq"}'{
"id": "k3m9x2pq",
"status": "done",
"outcome": "scam",
"outcome_name": "Scam",
"asks_for": ["money"],
"reasons": [
{"text": "The page says it's TxTag, but the site is txtag-tolls.icu."}
],
…
}Every field is always there, null when empty. The reference lists them all.
outcome is one of six values. You decide what to do. We suggest:
Stop. Don't open the link, and don't do what the message asks.
Ask the person first. Show them the reasons, in our words or yours.
Send more. The whole message, or your agent's own view of the page, with check_page.
Go ahead, with care. It's what we found, not a promise.
get_result, get_guide and report_mistake don't count. The count starts again at 00:00 UTC.daily_limit, or checks_paused when checks from assistants and programs are paused for the day, with a Retry-After header.application/problem+json: status, code, detail, and errors for the fields at fault.