Clawrouter

POSThttps://your-instance.com/v1/images/edits

Takes multipart form data rather than JSON. Supply the source image and a prompt describing the change; supply a mask as well to confine the edit to part of the frame.

modelstringrequired
Image model to use.
imagefilerequired
The image to edit.
promptstringrequired
What the edited image should look like.
maskfile
Transparent areas mark where the edit applies. Must match the source dimensions.
nintegerdefault: 1
How many variations to produce.
sizestringdefault: "1024x1024"
Output dimensions.

curl

curl https://your-instance.com/v1/images/edits \
  -H "Authorization: Bearer $CLAWROUTER_API_KEY" \
  -F model="gpt-image-2" \
  -F image="@room.png" \
  -F mask="@wall-mask.png" \
  -F prompt="Repaint the wall deep green"