Security fixes land on the most recent stable line. Earlier lines are supported at the maintainer's discretion and are marked "End-of-Life" once they stop receiving fixes. Pre-1.0 releases are alpha software and receive nothing.

Please do not report security issues publicly. Report them privately through either:

• GitHub Private Vulnerability Reporting: the Security tab, "Report a vulnerability".
• Email to asgerjon2@gmail.com with markwork security in the subject.

A report is most useful with a description of the issue, the steps to reproduce it, the affected version, and the impact.

markwork is maintained by one person, so there is no guaranteed response time. Reports are taken seriously and acknowledged as soon as is practical. Please allow a reasonable period for a fix before any public disclosure.

When a report comes in, the issue is confirmed, fixed, and released as a patched version on the supported line. The reporter is told when the fix is out and is credited for it unless they would rather stay anonymous. Please keep the details private until the fix is released, so that people can update before the issue is public.

Good-faith research and responsible disclosure are welcome. A reporter who acts in good faith, gives a reasonable chance to fix the issue before going public, and does not exploit it beyond what is needed to show that it exists, has nothing to fear from the project.

markwork runs at documentation build time. It reads a project's source files as text, parses them with ast, tokenises them with Pygments, and writes HTML pages; it opens no sockets, starts no processes, and never imports or executes the code it documents. Its real surface is therefore the generated HTML. All rendered source content is escaped with html.escape before it reaches the page, so a cross-site scripting issue in the published docs, where some token text escaped into live markup, is the thing to watch and the kind of report most in scope.

Out of scope is anything that requires a hostile build environment: a project that feeds markwork source it does not trust, or a docs toolchain already under an attacker's control, is outside what the build step can defend against.