Method and reference values

Reference values

The "Reference value" column on the result sheet shows what full marks require. A result below the reference gets the flag L; a result below 50 gets the flag ! (critical).

Scope

SecScan checks a host from outside, the way any browser or SSH client on the internet sees it. Checks are read-only: the scanner only sends protocol opening messages (TLS ClientHello, SSH banner) and reads the server's first reply. No login, no application data, and no attack attempts.

• A target is a domain or IP with an optional port (host:port, default 443). Up to 4 targets per check.
• Domains are resolved and the first address (IPv4 if there is one) is checked. Private, loopback and link-local addresses are refused, so the scanner cannot be used to reach internal networks.
• Only check systems you own or have permission to test.

How each probe works

TLS versions

For each version (1.0, 1.1, 1.2, 1.3) the scanner sends one ClientHello offering only that version. The version counts as enabled if the server answers with a ServerHello for the same version. The messages are built by hand, byte by byte, so the result does not depend on which TLS versions the scanning machine's library supports.

Post-quantum key exchange (TLS 1.3)

Each group is tested on its own. A TLS 1.3 ClientHello offers exactly one group in supported_groups with an empty key_share. A server that supports the group must answer with a HelloRetryRequest naming it (RFC 8446 §4.1.4); a server that doesn't answers with an alert. This way the scanner never needs its own ML-KEM implementation.

Certificate

The scanner makes an ordinary TLS handshake and reads the leaf certificate: public key type and size, signature algorithm, and expiry date. A second handshake validates the chain and host name against the system trust store. ML-DSA-44/65/87 and SLH-DSA OIDs are recognised as post-quantum certificates.

SSH

Ports 22 and 2222 are opened, the banner is read, and the scanner reads the server's KEXINIT message. That message is sent unencrypted before authentication and lists the key exchange and host key algorithms the server supports. The connection closes before any login.

How the score is calculated

Each check produces a score from 0 to 100. The composite score is the weighted average of the selected checks, rounded: score = Σ(weight × check score) / Σ weight. Checks that weren't selected don't count.

PQ key exchange

Certificate

Every classical key gets the same score. Shor's algorithm breaks RSA and ECC alike, and growing an RSA key to 3072 or 4096 bits adds no meaningful quantum resistance. Scoring big RSA keys higher would push migration in the wrong direction.

TLS versions

Score = points ÷ 40 × 100. For example: TLS 1.3 only = 100; TLS 1.2 + 1.3 = 62; no TLS 1.3 = 0.

SSH

For an open port, score = points ÷ 50 × 100.

Grades and PQC status

• PQC-ready: key exchange and certificate both score 100.
• Partly ready: key exchange is PQ, the certificate is still classical. This is the best a public site can reach today.
• Not ready: key exchange is not PQ yet.

Why these weights

• Key exchange (100) carries the largest weight because traffic recorded today can be opened once a quantum computer exists. Site owners can also fix it today with a software update.
• Certificate (60) protects the server's identity. Old signatures can't be forged after the fact, so the threat only becomes real once a quantum computer exists. Site owners can't do much about it until CAs issue PQC certificates.
• SSH (50) is the administration door. An SSH port open to the internet widens the attack surface, so keeping it closed scores highest.
• TLS versions (40) are a prerequisite: PQ key exchange only exists in TLS 1.3, and a TLS 1.2 that's still enabled lets clients fall back to classical key exchange.

The threat itself is explained in Understanding your result.

Limits

• Only one IP address per target is checked. Hosts with several IPs or a load balancer may give different results per node.
• Sites behind a CDN or reverse proxy are scored on the edge, not the origin server.
• Only the public surface is checked: HTTPS and SSH. VPNs, internal networks, application tokens (JWT), stored data, HSMs and firmware signing are invisible from outside and need their own cryptographic inventory (CBOM).
• Only the leaf certificate is scored; intermediate and root CAs are not.
• Firewalls or rate limits can block probes and make a feature look disabled.
• A result is a snapshot of the moment it was taken.