Free API

IABIG's tools are available as JSON, with no key and no sign-up, for your scripts, pipelines and agents. Add "lang": "en" to any request to get English messages.

Limits

20 URL checks per hour per IP address (checker and detector combined), 5 llms.txt generations per hour;
200 text requests per hour per IP address (injection text, robots.txt tester, IP verification), 256 KB maximum per request;
URL checks are cached for one hour ("cached": true);
only public addresses over http/https on ports 80 and 443 are fetched.

Errors look like {"error": "message"} with HTTP status 400 (bad request), 413 (too large), 422 (site unreachable or address refused) or 429 (rate limit).

AI visibility checker

curl -s https://iabig.com/api/v1/scan \
  -H 'Content-Type: application/json' \
  -d '{"url": "example.com", "lang": "en"}'

Returns score (0 to 100), grade (A to E), bots[] (token, company, kind = search, user or training, allowed, rule), sections[].checks[] (status = ok, warn, fail, info), injection and facts.

Prompt injection detector

import json, urllib.request

def check_injection(text):
    req = urllib.request.Request(
        "https://iabig.com/api/v1/injection",
        data=json.dumps({"text": text, "lang": "en"}).encode(),
        headers={"Content-Type": "application/json"},
    )
    with urllib.request.urlopen(req, timeout=10) as r:
        return json.load(r)

result = check_injection(document)
if result["risk"] == "élevé":  # stable codes: aucun, faible, à vérifier, élevé
    ...  # drop or sanitize the document

Send {"url": "…"} instead of text to check a web page. Each finding has severity (stable codes: critique, élevé, moyen, info), kind, where and snippet. Submitted text is never stored.

robots.txt tester

curl -s https://iabig.com/api/v1/robots-test \
  -H 'Content-Type: application/json' \
  -d '{"robots": "User-agent: GPTBot\nDisallow: /", "path": "/blog/"}'

Crawler IP verification

curl -s https://iabig.com/api/v1/verify-ip \
  -H 'Content-Type: application/json' \
  -d '{"ip": "203.0.113.10", "user_agent": "Mozilla/5.0 (compatible; GPTBot/1.2)", "lang": "en"}'

verdict is one of the stable codes authentique (genuine), usurpé (spoofed), non vérifiable (unverifiable), connu (in an official list, no User-Agent given) or inconnu (unknown); message explains it in plain English.

llms.txt generator

curl -s https://iabig.com/api/v1/llms-txt \
  -H 'Content-Type: application/json' \
  -d '{"url": "example.com", "lang": "en"}'

Observatory statistics

curl -s https://iabig.com/api/v1/stats

Verified AI crawler visits over the last 30 days and aggregated statistics of scanned sites. Data under CC BY 4.0: credit “IABIG Observatory” with a link.

Fair use

The API is free and provided without any availability guarantee. Only check sites you are allowed to test and respect the limits.