This is a small release focused on making Scaffold quieter, clearer, and easier for AI agents to use during document review.
A More Focused Pilot Surface
Templates and Sets are now hidden from the active web app and MCP pilot surfaces.
They have not been deleted. The underlying data, APIs, and historical implementation work remain in place, but the product now keeps the pilot focused on the workflows people are actively using: upload, redline, compare, review, version copy, download, and email.
Better Redline Formatting Context
Redline tools now give agents a clearer, bounded preview of what changed before the document reaches you for review.
Successful edits can return nearby before-and-after text, visible line breaks and tabs, compact run-formatting maps, relevant paragraph XML, and warnings for mixed-format paragraphs or bold-label/body-text issues. Batch edits also handle multiple replacements in the same paragraph more predictably, so one fix is less likely to flatten formatting from another.
The goal is simple: agents get enough context to correct formatting problems without flooding the conversation with the whole document.
Clearer Compare Workflows
Compare language is clearer in the review pane and agent tools.
Instead of vague "Base" and "Compare" labels, Scaffold now describes the two sides as the version to update and the version with changes. Actions are clearer too: add selected changes to the version you are updating, or revert selected changes in the changed version.
We also tightened the agent-facing Compare responses so normal compare calls return paginated, text-focused differences instead of large binary render artifacts. The full visual artifact still loads when you open or refresh the interactive Compare view, but routine agent work should use less context.
Small Session Update
Browser sessions now last 7 days by default, instead of expiring after the shorter previous window.
Questions or pilot feedback? Reach out at drew@scaffoldyourdocs.com.