farscry annotate

Draw bounding boxes over every detected element. Returns the original image with colored overlays.

Usage

farscry annotate screenshot.png
farscry annotate screenshot.png -o annotated.png
farscry annotate --from-clipboard -o /tmp/out.png

Color coding

Element type	Color	Hex
button, link	indigo	#6366f1
input, textarea, select	cyan	#22d3ee
error, alert	red	#f87171
heading	amber	#fbbf24
label, badge, other	gray	#a1a1aa

Affordances (clickable or typeable elements) get a thicker border (3px vs 2px).

Zero-friction alias

Add to ~/.zshrc:

echo "alias fannot='farscry annotate --from-clipboard -o /tmp/farscry_annotated.png && open /tmp/farscry_annotated.png'" >> ~/.zshrc && source ~/.zshrc

Then: Shottr screenshot -> fannot -> annotated image opens automatically.

Flags

Flag	Description
--from-clipboard	Read image from clipboard instead of file (macOS and Linux only)
-o, --output FILE	Output path (default: {input}_annotated.png)

Use cases

Verify element detection before automation: Run farscry annotate on a screenshot before wiring your agent. Confirm every button and input has a bounding box.

Debug agent failures: When your agent clicks the wrong element, annotate the screenshot to see what farscry detected vs what the agent targeted.

Share with your team: Annotated screenshots are immediately readable by non-technical stakeholders. No code required to understand what farscry sees.