> That plan sounds good.  What can I do to facilitate it?

I'll try to polish the code gradually, using the SRFI Explorations
GitHub organization as the testbed. Getting the automatic pull request
status checks working would be first priority.

Beyond that, I don't think it makes sense to do a lot of experimental
edits to the HTML yet because we don't know what kidn of final form it
will take.

Once we decide on the final form, I'll update the extraction and
validation code and we can mark up some SRFIs under SRFI Explorations
for assurance that everything works. Once we're reasonably certain, we
can install a production copy of the CI server and enable the webhook on
the real SRFI GitHub organization, then start marking up the SRFIs
there. At this point the automatic pull request checks and the visual
dashboard on the CI server's front page should be reliable indicators of
valid/invalid/missing metadata so we can use those to guide our work.

The main thing that can't be auto-extracted from the SRFI HTML is the
See Also links. Those would go in the manually edited metadata files.
Maybe we should commit those a bit later since the Scheme tool doesn't
yet read them (only the Python version does). If you have a reasonably
comprehensive list of the See Also for each SRFI, we could transform
that into the manual metadata files and then commit those, but that can
be done whenever it's convenient to you and they can be edited/fixed
later at any time.

Thanks for the great support :)