v0.28.0
June 22, 2026
- Building a premise by hand is now direct: instead of choosing "Add first claim" or "Add inference" up front, an empty premise shows fill-in "skeleton" nodes you complete by tapping a "+". Selecting a claim offers to extend just that copy in place, and after you add a claim it stays selected so you can keep building the rest of the premise. You can turn a single-claim premise into an inference — and back, with a small cancel on the placeholder — without deleting and re-adding claims.
- Editing an argument now gives clearer feedback while changes save: changing a logic operator, adding a source, assigning or removing a justification, and saving a new source all show a spinner and disable their button until the change goes through, so you no longer wonder whether a slow action registered. You can also always add another source to a claim that already has one, and a couple of claim menu labels are clearer ("Negate this claim", "Remove").
- Clicking a node in an argument now selects exactly what you clicked and scrolls right to it, instead of jumping to a different copy of the same claim. Shared links that point at a node land precisely on it, and if a link points at something that can't be selected you'll get a clear message instead of nothing happening.
- Fixed: if the AI assistant became temporarily unavailable during an import, the Argument Builder could stay hidden on that draft. The builder now stays available beneath a short notice, so you can keep editing or try again once the assistant is back.
- Only an argument's owner can delete an unpublished draft now. Collaborators with edit access keep full editing rights but can no longer delete the draft.