canary
How likely is a phishing kit's cloaking layer to classify this traffic as a scanner and hide the payload from you?
165/ risk points
critical
You will be cloaked. Analysis results are not trustworthy.
Running browser-side checks…
Server-side decided before any HTML is sent
info
TLS fingerprint (JA3)
This hash identifies your TLS stack. Kits and CDNs match it against known-automation lists; it is stable across runs and independent of anything you set in HTTP.
JA3=7c4f2964e18b6f743c8ec9102a0c3210 negotiated=TLSv1.3 TLS_AES_256_GCM_SHA384 alpn=http/1.1
flag
TLS GREASE+35
The ClientHello contains no GREASE values. Every real Chrome and Firefox inserts them; Python, Go, curl and most automation stacks never do. With a browser user-agent this is conclusive proof the traffic is not from the browser it claims to be.
no GREASE in cipher suites or extensions
ok
TLS 1.3 support
TLS 1.3 offered.
ok
ALPN
Offers h2, as a browser does.
h2, http/1.1
ok
Cipher suite list
10 cipher suites offered.
ok
SNI
Matches the Host header.
canary.randomwisman.com
flag
User-Agent header+35
UA self-identifies as tooling ('Bot'). This alone is a block on most kits.
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; ClaudeBot/1.0; +claudebot@anthropic.com)
info
Sec-CH-UA client hints
Not applicable for this UA.
flag
Sec-Fetch-* metadata+22
No Sec-Fetch-Site/Mode/Dest/User. Every browser since 2020 sends these on a top-level navigation.
ok
Accept header
Plausible.
*/*
ok
Accept-Encoding
Plausible.
gzip, br, zstd, deflate
flag
Accept-Language+15
No Accept-Language. Kits geo-gate on this and block when absent.
flag
Header order+18
Only 33% of Chrome's header orderings preserved. Header order is a stable client fingerprint and is cheap for a kit to check.
accept -> user-agent -> accept-encoding -> host
ok
Proxy / scanner headers
No proxy or scanner headers leaked.
info
HTTP version
HTTP/1.1. (This server is h1-only, so h2/h3 negotiation and JA3/JA4 TLS fingerprinting are not tested here -- see the notes below.)
HTTP/1.1
warn
Referer+10
No Referer. Many kits only serve the payload when the referrer is a mail provider or link shortener, and show a decoy otherwise.
flag
Egress IP / ASN+30
Egress is a hosting/datacenter ASN. Residential victims never originate here, and datacenter ASN lists are the first filter in nearly every kit. This is the highest-value thing to fix.
216.73.217.68 AS16509 AMAZON-02 - Amazon.com, Inc., US rDNS=none
info
Cookie persistence
First request -- reload this page to test whether cookies persist.
Browser-side decided in JS, after the page loads
JavaScript did not run. A kit that gates on JS would show you a decoy page and you would never know. If your tool renders JS, this section should fill in within a few seconds.
Plain-text report for tools that capture text, not DOM
CANARY -- CLOAKING EXPOSURE CHECK
time: 2026-07-21T19:19:02.351308+00:00
client: 216.73.217.68 GET / HTTP/1.1
SERVER-SIDE
[INFO] TLS fingerprint (JA3)
This hash identifies your TLS stack. Kits and CDNs match it against known-automation lists; it is stable across runs and independent of anything you set in HTTP.
> JA3=7c4f2964e18b6f743c8ec9102a0c3210 negotiated=TLSv1.3 TLS_AES_256_GCM_SHA384 alpn=http/1.1
[FLAG] TLS GREASE +35
The ClientHello contains no GREASE values. Every real Chrome and Firefox inserts them; Python, Go, curl and most automation stacks never do. With a browser user-agent this is conclusive proof the traffic is not from the browser it claims to be.
> no GREASE in cipher suites or extensions
[OK ] TLS 1.3 support
TLS 1.3 offered.
[OK ] ALPN
Offers h2, as a browser does.
> h2, http/1.1
[OK ] Cipher suite list
10 cipher suites offered.
[OK ] SNI
Matches the Host header.
> canary.randomwisman.com
[FLAG] User-Agent header +35
UA self-identifies as tooling ('Bot'). This alone is a block on most kits.
> Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; ClaudeBot/1.0; +claudebot@anthropic.com)
[INFO] Sec-CH-UA client hints
Not applicable for this UA.
[FLAG] Sec-Fetch-* metadata +22
No Sec-Fetch-Site/Mode/Dest/User. Every browser since 2020 sends these on a top-level navigation.
[OK ] Accept header
Plausible.
> */*
[OK ] Accept-Encoding
Plausible.
> gzip, br, zstd, deflate
[FLAG] Accept-Language +15
No Accept-Language. Kits geo-gate on this and block when absent.
[FLAG] Header order +18
Only 33% of Chrome's header orderings preserved. Header order is a stable client fingerprint and is cheap for a kit to check.
> accept -> user-agent -> accept-encoding -> host
[OK ] Proxy / scanner headers
No proxy or scanner headers leaked.
[INFO] HTTP version
HTTP/1.1. (This server is h1-only, so h2/h3 negotiation and JA3/JA4 TLS fingerprinting are not tested here -- see the notes below.)
> HTTP/1.1
[WARN] Referer +10
No Referer. Many kits only serve the payload when the referrer is a mail provider or link shortener, and show a decoy otherwise.
[FLAG] Egress IP / ASN +30
Egress is a hosting/datacenter ASN. Residential victims never originate here, and datacenter ASN lists are the first filter in nearly every kit. This is the highest-value thing to fix.
> 216.73.217.68 AS16509 AMAZON-02 - Amazon.com, Inc., US rDNS=none
[INFO] Cookie persistence
First request -- reload this page to test whether cookies persist.
SERVER SCORE: 165 (critical) -- You will be cloaked. Analysis results are not trustworthy.
RAW REQUEST HEADERS (in order received)
accept: */*
user-agent: Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; ClaudeBot/1.0; +claudebot@anthropic.com)
accept-encoding: gzip, br, zstd, deflate
host: canary.randomwisman.com
What this does not test
- JA4. JA3 is computed; the newer JA4 suite is not. They fingerprint the same handshake, so a JA3 that looks like Chrome generally implies a JA4 that does too.
- HTTP/2 fingerprint. Frame settings and pseudo-header order are a second fingerprint that HTTP/1.1 cannot exercise. Chrome will fall back to h1 here because this server does not advertise h2 in ALPN.
- ASN reputation. Being on a residential ASN is not enough; kits also buy known-proxy-exit lists. A clean ASN result here is necessary, not sufficient.
Endpoints: /report.json full machine-readable
result · /plain no-JS text report ·
/traps what redirect/robots traps your tool followed.