diff --git a/.github/workflows/markdown-linting.yml b/.github/workflows/markdown-linting.yml
index 173f461d29..11567d8c36 100644
--- a/.github/workflows/markdown-linting.yml
+++ b/.github/workflows/markdown-linting.yml
@@ -14,5 +14,4 @@ jobs:
       - name: MarkdownLint mdl Action
         uses: actionshub/markdownlint@v2.1.2
         with:
-          # filter out version-specific/* for now, until we fix all issues that mdl trips over
-          filesToIgnoreRegex: "(docs\\/demos\\/.*|docs\\/version-specific\\/.*)"
+          filesToIgnoreRegex: "(docs\\/demos\\/.*|docs\\/version-specific\\/(supported-software|generic-easyblocks).md)"
diff --git a/docs/controlling-compiler-optimization-flags.md b/docs/controlling-compiler-optimization-flags.md
index 5c71f92a74..ec3b4a3d58 100644
--- a/docs/controlling-compiler-optimization-flags.md
+++ b/docs/controlling-compiler-optimization-flags.md
@@ -37,7 +37,7 @@ it will not run on a system with an older Intel Nehalem processor.
 One possible workaround for heterogeneous HPC clusters is to build as many copies of the software stack as you have
 processor generations in your cluster, and to configure your system so each compute node uses the right software stack
 matching its processor architecture type.
-Details for one way of doing this, using automounter/autofs are available at <https://easybuilders.github.io/easybuild/files/sciCORE-software-management_20150611.pdf>.
+Details for one way of doing this, using automounter/autofs are available at <https://easybuild.io/files/sciCORE-software-management_20150611.pdf>.
 
 Another solution is to configure EasyBuild to not optimize for the processor architecture of the build host via
 `--optarch`, see below.