Being a security tool, it should obviously be the goal that badkeys itself is as secure as possible.

If you have found a severe security bug that you want to report confidentially, please contact Hanno Böck directly.

Low-impact security issues can be directly reported to the badkeys issue tracker.

Given that badkeys is a software that reads and processes cryptographic keys in various formats, the most likely attack surface for any security issues is the processing of untrusted inputs.

badkeys should be able to handle malformed and malicious input files. The following types of issues would be severe security vulnerabilities:

• A malicious input can trigger the execution of malicious code.

• A malicious input can trigger any kind of network request.

• A malicious input can exfiltrate information about the system via the program's output.

Malformed input files that cannot be scanned should cause a warning, but no abnormal application behavior. The following types of issues would be treated as denial of service issues that should be fixed, but their security impact is less severe:

• A malformed input causes an unexpected termination of the badkeys application, e.g., by causing an OOM condition, a crash, or an unexpected exception. (Note, however, the caveat below regarding dependencies.)

• A malformed input causes an endless loop, an excessively long processing time, or excessive resource consumption.

The blocklist functionality requires downloading a list of hashes of known-compromised keys and optionally a list of URLs via the --update-bl or --update-bl-and-urls parameters. This fetches the latest blocklist updates via HTTPS from update.badkeys.info.

The connection is TLS-protected. A network attacker would naturally be able to prevent an update from succeeding. However, any issue where a network attacker can tamper with the connection in a way that causes badkeys to download and use a manipulated blocklist update would be a severe security flaw.

Another potential scenario would be a compromised server or web certificate. If an attacker is able to serve TLS-protected content to badkeys, it is expected that he could serve a malformed blocklist that will not function correctly and may terminate unexpectedly. However, a malformed blocklist should not cause malicious code execution, network requests, or data exfiltration. Any such issue would also be treated as a security vulnerability.

The security of a software relies to a certain degree on its infrastructure. Reports about security issues of the badkeys.info web page (source code), the server where it is hosted, the code repositories, or the pypi packages are also appreciated and will be addressed accordingly.

Lists of potential security issues are always incomplete, as the most severe and interesting security issues are often the ones one has not thought about.

badkeys relies on functionality provided by dependencies. Naturally, this means that security issues and bugs in dependencies can cause problems in badkeys.

Obviously, bugs and security issues in dependencies should be fixed. However, badkeys will usually not implement workarounds to address bugs in other software.

An example: key parsing by badkeys relies on the Python cryptography package. During the development of badkeys and its supporting tools, bugs in cryptography's key parsers were uncovered that could cause unexpected exceptions. Such bugs can cause an abnormal program termination of badkeys when using an outdated version of cryptography. That is, however, not treated as a bug in badkeys.

It is recommended to run badkeys with the latest versions of its dependencies. Currently, all known parser issues in cryptography have been addressed in version 46.0.3.

Security issues or parser bugs in dependencies should be reported to those projects directly. However, if you discover any bugs in dependencies during the use of badkeys, I would appreciate a note.