What the connector exposes

Everything an AI assistant can read or change through the connector, by area.

The connector exposes a little over a hundred operations, grouped by what they act on. Roughly two fifths only read; the rest can change something, and those are the ones the write permission gates.

By area

AreaCovers
PagesThe biggest surface by far: listing, rotating, clipping, renaming, renumbering, importing, placing, comparing, scale, titles and page context.
UtilitiesSystems, stages, pipes, structures, runs and references.
MaterialsItems, sub-items, layers and quantities.
MarkupListing, creating, updating and deleting annotations, and holes.
Groups and collectionsPage groups and collections: creating, assigning, merging.
ProjectThe project map and summary, plus a project memory it can write notes into.
DocumentsReading documents: outline, sections, requirements, and querying them.
SpatialMeasuring, resolving a sheet, snapping a line, and the vectors inside a region.
Vectors and textThe drawing's own geometry and text: searching, extracting a region, layer lists.
ScaleSetting, calibrating and auditing page scales.
ViewNavigating and capturing the canvas.

What that is good for

The useful question is not which tools exist but what they add up to:

The limits worth knowing