Use TagDraft when someone who doesn't write HTML needs to keep a plain semantic HTML site up to date: they open the page as visitors see it, click the words, the photo, or the table row they want to change, and save.
Status: prototype. There is a throwaway prototype to try the idea with, at kovirlabs.github.io/tagdraft. It can't save to a site yet. Spec.md says what the real editor will be and lists the questions still open.
TagDraft is part of the Tag family. It edits pages styled by TagBrush.css, made interactive by TagAction, checked by TagCheck, and assembled by TagKit.
A site written in plain semantic HTML needs no CMS, database, or framework, and it will still open in twenty years. The cost is that every change means editing HTML by hand. The usual fix is to move the content into Markdown or YAML and add a site generator and a CMS on top, which brings back the moving parts the plain site avoided.
TagDraft takes the other route: keep the HTML as the only source, and make the HTML editable. Because Tag family pages are classless and semantic, an editor can tell what something is from its tag alone. A <table> is rows to add and remove, a <time> is a date to pick, an <article> in a <section> is a card to copy. No schema or configuration file has to describe the content first.
The church starter in TagKit: a 14-page site whose weekly changes (this Sunday's sermon, the events table, a new sermon page) are marked with comments. TagDraft is ready when a church volunteer can make those changes without seeing a tag.
prototype/ is a static page: serve the folder (python3 -m http.server 8791 inside it) and open it. It edits a built copy of the church starter in prototype/site/, adds sections from TagKit's snippets, and checks pages with TagCheck's static rules, both copied into prototype/vendor/. prototype/setup.sh refreshes those copies from tagkit and tagcheck checked out beside this repository.
Apache-2.0.