Previously, Puppet Discovery was shipped with a default generated TLS certificate in the nginx container. In version 1.4.0, a unique certificate will be generated on installation or the user will be able to provide their own TLS certificate for ingress.

Previous releases of the Puppet cisco_ios module output SSH session debug information including login credentials to a world readable file on every run. These issues have been resolved in the 0.4.0 release.

Previous releases of the Puppet cisco_ios module did not validate a host's identity before starting a SSH connection. As of the 0.4.0 release of cisco_ios, host key checking is enabled by default.

Previous releases of the Puppet device_manager module creates configuration files containing credentials that are world readable. This issue has been resolved as of device_manager 2.7.0.

When users are configured to use startTLS with RBAC LDAP, at login time, the user's credentials are sent via plaintext to the LDAP server. This affects Puppet Enterprise 2018.1.3, 2017.3.9, and 2016.4.14, and is fixed in Puppet Enterprise 2018.1.4, 2017.3.10, and 2016.4.15. It scored an 8.5 CVSS score.

In Puppet Discovery prior to 1.2.0, when running Discovery against Windows hosts, WinRM connections can fall back to using basic auth over insecure channels if a HTTPS server is not available. This can expose the login credentials being used by Puppet Discovery.

On Windows only, with a specifically crafted configuration file an attacker could get Puppet PE client tools (aka pe-client-tools) 16.4.x prior to 16.4.6, 17.3.x prior to 17.3.6, and 18.1.x prior to 18.1.2 to load arbitrary code with privilege escalation.

The previous version of Puppet Enterprise 2018.1 is vulnerable to unsafe code execution when upgrading pe-razor-server. Affected releases are Puppet Enterprise: 2018.1.x versions prior to 2018.1.1 and razor-server and pe-razor-server prior to 1.9.0.0.

Puppet Enterprise 2016.4.x prior to 2016.4.12, Puppet Enterprise 2017.3.x prior to 2017.3.7, Puppet Enterprise 2018.1.x prior to 2018.1.1, Puppet Agent 1.10.x prior to 1.10.13, Puppet Agent 5.3.x prior to 5.3.7, and Puppet Agent 5.5.x prior to 5.5.2, were vulnerable to an attack where an unprivileged user on Windows agents could write custom facts that can escalate privileges on the next puppet run. This was possible through the loading of shared libraries from untrusted paths.

Puppet Agent 1.10.x prior to 1.10.13, Puppet Agent 5.3.x prior to 5.3.7, and Puppet Agent 5.5.x prior to 5.5.2 on Windows only, with a specially crafted configuration file an attacker could get pxp-agent to load arbitrary code with privilege escalation.