1.6.3
Released 24 Oct 2026
New
-
Discovery finding notifications. Subscribe a rule to Discovery: Finding Opened to hear when a finding opens, comes back after it was resolved, or becomes more severe. More than five at once arrive as one summary per discovery source.
-
Expiry alerts for discovered certificates. Discovery: Certificate Expiring uses the same lead times as managed certificates (30, 14, 7 and 1 days by default, then daily in the last 3 days), Discovery: Certificate Expired is sent once, and more than five at once arrive as one summary per discovery source. Certificates CertAutoPilot manages are left out because their own expiry alerts already cover them.
-
Discovery verifies certificate chains. Each endpoint's chain is checked against the system roots and an optional CA certificate on the source, and new findings report an untrusted chain, a missing intermediate, a chain out of order, a revoked certificate, a CA certificate served as a server certificate, a key usage that excludes TLS servers, and a public certificate valid longer than the CA/Browser Forum limit.
-
Discovery reads stapled OCSP responses and can ask the CA about revocation. The stapled response is checked on every scan; with Ask the CA about revocation on, each certificate is also checked with its CA's OCSP responder or CRL at most once a day.
-
STARTTLS services are discovered. SMTP (25, 587), POP3 (110), IMAP (143), LDAP (389), PostgreSQL (5432), RDP (3389), XMPP (5222, and 5269 between servers) and FTP (21) are scanned through their plain-text upgrade, and any other port can be marked with its protocol, for example
2525:smtp. -
Fetch missing intermediates. With Fetch missing intermediates on, a source downloads the intermediate a certificate's issuer URL names when its server sends an incomplete chain, shows whether that completes the chain, and names the certificate to add in the finding.
-
IPv6 discovery. CIDR and IP records, zone transfers and cloud-DNS sources now include IPv6 addresses; unique-local addresses count as private.
-
Scan every address of a hostname. A hostname source can scan each address the name resolves to as its own endpoint, so every backend behind a round-robin name is checked.
-
CAA and HSTS checks. Publicly trusted certificates are checked against the CAA records of their names, and with Check HSTS on each named HTTPS endpoint's Strict-Transport-Security header is read.
-
Discovery suggests distribution targets. Discovery → Target candidates lists the hosts where discovery found a real certificate, with a type suggestion where the scan shows one, and Add as target opens the target form with the address and validation endpoints filled in. A removed candidate never comes back on a later scan and can be restored or still added from the Removed list.
-
TLS-settings changes are recorded. The change history of a discovered certificate now also shows a lower TLS version, a weaker cipher, a lost post-quantum key exchange or a chain that is no longer trusted while the certificate stayed the same, and a change for the worse is notified.
-
Filter, sort and export the discovery inventory. Certificates can be filtered by issuer, key type, managed or not, risk, PQC status, source, tag and owner and sorted by risk, and both the inventory and the findings export to CSV or JSON.
-
Tags and an owner on discovered certificates. They stay with the certificate even when discovery stops seeing it, and the inventory and the findings list filter by them.
-
Change many findings at once, and reopen them. Tick findings, or select every finding matching the filters, to acknowledge, suppress, resolve or reopen them in one step; a single finding can now be reopened too.
-
Import a discovered certificate with only its private key. Discovery now keeps the certificate and chain each server presents, so a project admin can import it as a managed certificate by supplying the key.
-
The Manage wizard can issue from Microsoft AD CS and shows the places the certificate was seen that can become distribution targets.
Improved
-
Security Findings name the certificate. A certificate finding shows the certificate's name and a server it was seen on under its title, so findings that share a title can be told apart.
-
A revoked certificate looks revoked. On a discovered certificate the validity bar turns red and reads Revoked instead of counting down the days left, and on discovered and managed certificates alike the chain card marks the certificate Revoked.
-
Discovery explains its labels. Every discovery page has a Labels explained panel describing each tag it shows, and the tags that say how a name was found (SAN crawl, Candidate pool and the others) explain themselves on hover.
-
Findings sort by severity properly. Critical comes first, then high, medium, low and info; before, the list sorted the names alphabetically and put info before low and medium.
-
A certificate seen on very many endpoints opens quickly. Its page lists the first 5,000 endpoints and says how many there are.
-
Accepted weak cipher suites are grouped. The endpoint row shows one chip per kind and TLS version with a count, and the chip's tooltip names every suite.
Changed behaviour
-
Discovery findings are recomputed every hour, not only after a scan, so an expiry window crossed between two scans shows up within the hour.
-
A finding that closed by itself less than a day ago and comes back reopens without a notification, so a check that flaps does not announce every return.
-
Rules saved by users limited to granted resources now receive discovery events for the discovery sources they are granted. Before, no discovery event reached such a rule.
-
A name answered with a certificate that does not cover it is now a finding (hostname mismatch) when DNS points the name at that server. Before it was only a flag on the endpoint.
-
Ports 21, 25, 587, 110, 143, 389, 5222, 5269, 5432 and 3389 are scanned through STARTTLS. A source that already lists them starts finding their certificates on the next scan instead of failing; write
port:tlsto keep TLS from the first byte. -
Loopback, link-local, multicast and unspecified addresses are refused when a discovery source is saved, whatever the public-scan setting says. They were never scanned.
-
A suppressed finding stays suppressed. Discovery no longer closes it and opens it again when the certificate or the problem goes away and comes back; only Reopen brings it back.
-
Finding status changes follow one rule. For example an acknowledged finding cannot be acknowledged again and an automatically resolved one cannot be changed by hand; a change the current status does not allow is refused.
-
The inventory lists the risk level stored after each findings refresh, so the list can sort and filter by it; its expiry-based part can be up to about an hour behind, while the certificate's own page computes it fresh.
-
A user limited to granted discovery sources gets an error instead of an empty list when what those sources found is too large to work out (about 190,000 certificates or more); grant fewer sources or the whole project.
-
A certificate is self-signed only when its own key signed it. A certificate that merely names itself as issuer no longer gets a Self-signed finding.
Scan every address can only be turned on once every node runs 1.6.3; until then saving it is refused. An older worker scans STARTTLS ports, IPv6 addresses and chains as before and records XMPP and FTP ports as failed. It still closes suppressed findings whose certificate went away — such a finding comes back as open on a 1.6.3 node, so check suppressed findings after the upgrade. It fetches no missing intermediates and does not store certificates for import or risk levels for the inventory (until a 1.6.3 node recomputes the findings, the inventory's critical and high counts can read zero).
A rule subscribed to Discovery: Finding Opened receives one summary per discovery source of the findings that are already open, and Discovery: Certificate Expiring sends one summary per source for the certificates already within a lead time. After that only changes are reported.
Fixed
-
Viewers see the endpoints of a discovered certificate. The certificate page loaded them through a request only operators may make, so the table stayed empty for viewers.
-
A certificate CertAutoPilot manages no longer gets an Unmanaged certificate expires finding when discovery sees it.
-
The PQC-vulnerable key finding opens again for certificates with a vulnerable key, and finding descriptions name the certificate's subject again.
-
Deleting a discovery source as a user limited to granted sources no longer closes findings that only other sources see.
-
Hostname discovery sources use the Discovery DNS Resolvers to connect. A name only an internal resolver knows used to fail with "no such host" before the scan started.
-
OCSP responses signed by the issuing CA itself are accepted when the response also carries the CA's own certificate. Such valid responses used to be refused, for managed-certificate revocation checks too.
-
LDAP sign-in works with an IPv6 address as the server, entered with or without brackets.
-
Importing a discovered certificate signed with SHA-1 keeps its chain. The intermediates it was imported with were left out, so the imported certificate was distributed without them.
-
Expanding a finding on the Security Findings page shows its description and how to fix it. Both were stored but never reached the page.
-
A discovery source's endpoint totals say "hosts", not "IPs", since a hostname source groups by name; and the Add Source form opens at its top again instead of where it was last scrolled.
-
The Vulnerable quantum-readiness label no longer says RSA 2048 is vulnerable. Only RSA keys shorter than 2048 bits are rated Vulnerable, as the classification always did.
-
Discovery change history entries name their endpoint. The endpoint of a recorded certificate change was left empty.