Grouped by what each tool operates on. Every card opens its own page with the inputs for that job.
Reads meaning out of an image. Model-backed, returns scores and text — never a verdict.

Extract text from posters, documents, scanned images, and PDFs.

Detect faces in an uploaded image.
Match uploaded faces against a named dataset.

Grade how exposed a subject is on a 0-3 ladder. Returns a probability per step plus a legacy blocked flag — threshold the scores yourself.
Pixels in, pixels out. Deterministic operations on the image itself.

Remove the background from a model or product image and return the cutout artifact.

Composite a subject image onto a new uploaded background with cutout, scale, anchor, and offset controls.

Score and cut out a fashion product image — detects model pose, removes background, and returns editorial suitability metadata.

Score how good a headshot a photo is — face angle, centring, size and resolution. Bulk-mode geometry only; no taste term.

Generate a BenaKino-style square headshot crop from an uploaded image.

Trim white borders using the BenaKino cover-crop heuristic.

Add same-size inset padding, an overlaid frame border, or both, using a chosen solid color.

Upload an image to extract the dominant hex color and the rest of its main palette.

Apply an uploaded watermark image onto a source image or PDF with opacity, size, position, and tiled placement controls.
Separate Option B background-removal pipeline scaffold for Florence-2, SAM2, and BiRefNet-style processing.
Separate Option B lifestyle composite path with advanced-stack runtime metadata and the same placement controls.
Structure and conversion — metadata, merging, and 1:1 metric DXF.
Never sees an image. Slugs, formatting, colour conversion and contact decoding.