Can I draft and edit CAD geometry inside a PDF editor?

A command line, not a toolbar hunt

The command line docks at the bottom of the workspace and works the way CAD operators expect: type a command or its alias — L for LINE, PL for PLINE, O for OFFSET, TR for TRIM — and follow the prompts. Autocomplete suggests commands as you type, multi-step commands collect points on the canvas with a rubber-band preview, Enter advances a step, and Esc cancels cleanly.

• Draw — LINE, PLINE, RECTANG, CIRCLE, ARC, ELLIPSE, HATCH, FREEHAND.
• Modify — MOVE, COPY, OFFSET, ROTATE, SCALE, TRIM, EXTEND, FILLET, JOIN, EXPLODE, ERASE, plus FENCE_TRIM and FENCE_EXTEND for cutting many objects with one line.
• Annotate — MTEXT, DDEDIT, DIMLINEAR, DIMALIGNED, LEADER, CLOUD, ARROW, PIN, HIGHLIGHT, COUNT, CALIBRATE.
• Navigate and manage — ZOOM, PAN, REGEN, UNDO/REDO, LAYER, COLOR, UNITS, PURGE, OVERKILL for duplicate cleanup.

The output is real, editable geometry

Commands do not burn pixels — they create the same annotation objects the markup tools produce. A polyline drawn with PLINE can be selected, moved, restyled, rotated with ROTATE, exploded with EXPLODE, or undone like anything else on the sheet. DIST and AREA report distances and surface areas, and DIMLINEAR/DIMALIGNED lay dimension linework on the page.

Snapping onto the drawing itself

OSNAP (F3) snaps your click to the PDF’s own vector linework — the closest point on the nearest segment within a zoom-scaled tolerance — so new geometry lands on real endpoints and edges instead of near-misses. Smart Ortho Snap straightens near-horizontal and near-vertical segments automatically while you draw.

Vectorizing the sheet and moving geometry out

CONVERTTOCAD (alias CTC or VECTORIZE) extracts the page’s native PDF linework, merges the raw segments into clean polylines in a background worker, and stores the result as a CAD layer you can select, trim, and erase like markup. Going the other way, the DXF command exports the current page’s geometry to a .dxf — imported linework on a PDF_IMPORT layer, your edits on USER_EDITS — and IMPORT:DWG converts a native DWG or DXF through the conversion service and opens it as its own tab.

The AI operator and the batch namespaces

The AI command takes a plain-language instruction — "move the selected rectangles 20 pixels right" — sends the selected geometry to the model, validates each returned edit, and applies the moves, deletions, and duplicates it proposes. Creating brand-new shapes from a sentence is the one thing it does not do yet; it edits geometry you point it at. The same line also runs set-wide jobs (FOLDER:SCALE_SYNC, FOLDER:SLIPSHEET, FOLDER:TAKEOFF_REPORT) and server-side document ops (CLOUD:OCR, CLOUD:TO-WORD), so the whole workspace is scriptable from one prompt.

Doing it in PDFDepth

Ctrl+9 opens the command line; type ? for the in-app command list. Draw with aliases, snap with F3, ortho with the Snaps toggle, then DXF the page out when the geometry needs to travel. Everything runs in the tab — the same session on a site tablet as on the office machine.