Is your certificate trusted everywhere it is served?
One scan reads the certificate chain, protocol support, ciphers and known vulnerabilities on every IP a host resolves to, and grades the result A+ to F.
Reports are public permalinks, not indexed by search engines. Domain owners can opt out on the opt-out page.
European by default. Scanning servers in Germany and Finland, reports stored in an EU database. How the data is handled.
Protocol support, cipher strength, key exchange and certificate quality are scored into a 0–100 total, following RFC 9325 (BCP 195). The worst finding then caps the total, so one serious fault decides the grade no matter how good the rest is. Every report names the finding doing the capping.
Scores inside a letter carry a plus or a minus, so 83 shows as B+ and 88 as A−.
What a single scan covers
Every scan runs the complete check set: no quick versus extended toggle, no option to skip a section.
How a scan runs
From submitted host to graded permalink, every scan takes the same five steps.
Built for automation
Run the same scan from a script, a CI pipeline or an AI agent. Free, no API key.
One GET, the whole report
The same graded data a browser sees, as JSON. One endpoint, live progress over SSE, no signup and no API key.
Read the API docs →A native tool for AI agents
Check TLS over the Model Context Protocol. Works with Claude, Cursor, VS Code and more. No key, no signup.
Read the MCP docs →Frequently asked
Yes. Every check, the API, and the MCP server are free with no account, signup, or API key. Usage is rate-limited for abuse prevention.
The full certificate chain, hostname match, expiry, key strength and permitted key usage, the certificate served without SNI, supported TLS versions, cipher suites and key-exchange groups, known vulnerabilities (Heartbleed, ROBOT, Ticketbleed, DROWN, POODLE, FREAK, Logjam and more), a client simulation across 160+ recorded browser and library handshakes, compromised-key lookup, HSTS, CAA, OCSP / CRL revocation status, and connection-security details like Extended Master Secret and post-quantum key exchange.
14 protocols: direct-TLS HTTPS, FTPS, SMTPS, POP3S, IMAPS, and LDAPS, plus STARTTLS on SMTP, SMTP submission, IMAP, POP3, FTP, LDAP, XMPP, and NNTP.
The grade summarises protocol support, cipher strength, key exchange and certificate quality, following the current TLS best practice in RFC 9325 (BCP 195, which obsoletes RFC 7525). Serious findings cap the grade regardless of the rest. What a standard prohibits outright caps at F: a revoked certificate, a key below 1024 bits, a certificate invalid for TLS server authentication, RC4 (RFC 7465) or a critical vulnerability. SSLv3 (RFC 7568) and other vulnerabilities cap at C, and deprecated TLS 1.0/1.1 (RFC 8996) or static RSA key exchange (RFC 10015) cap at B.
A report is kept as a shareable permalink for 24 hours. Use Re-scan on the report page for a fresh result at any time.
A padlock means one client reached one address and accepted the certificate. Browsers also repair a broken chain quietly: given a missing intermediate they fetch the missing certificate themselves, or reuse one cached from another site, and show a padlock over a server sending an incomplete chain. The grade describes the server, not one visit.
Almost always a missing intermediate certificate. curl, Python requests, Node, Go and Java build the chain only from what the server sends, so a gap browsers paper over becomes "unable to get local issuer certificate" or "PKIX path building failed". Older Android devices fail the same way. The fix is serving the full chain, which for most tooling means fullchain.pem rather than cert.pem.
The report answers directly. Every scan replays the ClientHello real crawlers send, recorded from live crawler traffic and refreshed as they change, so a host offering only ciphers Googlebot no longer accepts shows up as a failed client rather than as a mystery drop in crawling. Browsers and crawlers are listed separately, each with the TLS version and cipher negotiated.
Public TLS certificates issued since 15 March 2026 last at most 200 days, down from 398. The maximum drops to 100 days in March 2027 and 47 days in March 2029, and domain validation has to be repeated far more often alongside. Renewal still involving a person will not survive the schedule, so every report shows the days remaining and the API returns the same number for a cron job to watch.
For anything with a long confidentiality life, yes. Traffic recorded today stays readable to whoever kept it once a quantum computer breaks the elliptic-curve key exchange protecting the session, and hybrid key exchange closes it now rather than at some future upgrade. Current Chrome, Firefox and Safari already negotiate X25519MLKEM768 by default, so a server enabling the group starts using post-quantum key exchange for a large share of connections immediately. Every report names the group each client settled on.
Android 4.3 and older, Java 7 and older, OpenSSL 0.9.8 and Windows XP. Everything current speaks TLS 1.2, and TLS 1.3 covers the overwhelming majority of browser traffic. The client simulation in the report is the honest check: it names exactly which clients stop connecting, so the decision rests on the audience rather than on a guess.
A scan opens a small number of ordinary TLS handshakes, the same connections a browser or mail client makes, and reads what comes back. Nothing is written, no credentials are attempted, and no traffic is sent that a server is not designed to answer. Owners preferring no scans at all publish a DNS record on the opt-out page, and the host is refused from then on.