Cookie Cutter Press · مهارة Open Skill عامة
Cookie Cutter Press
Cookie Cutter Press هي Open Skill عامة على Digen. افتحها لتعرف ماذا ينشئ هذا التدفق.
Other
0 0 0
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/4284862/2026-10-16/scuu74eg.jpg
# Cookie Cutter Press Turn one uploaded reference image into a single square image: the subject's outer silhouette becomes a metal cookie cutter, sitting next to a same-shaped piece freshly pressed out of dough. This is image-to-image — the reference image is the only input, and the result is one still image. ## When to use - The user uploads one or more images and asks for a cookie-cutter / cookie-mold / press-cut version of the subject. - The user wants to redo it with a different photo, or to adjust an earlier result (cleaner outline, different dough tone, tighter framing) while keeping the same subject outline. - Do not use it for generic "draw me a cookie" or food-photography requests that have no reference subject. ## Steps 1. **Get the reference image.** Exactly one reference image is required. If none was uploaded, ask for one and stop. If several were uploaded, ask which single subject should be cut — one cutter, one closed outline. 2. **Name the subject and its signature outline features.** Briefly state what the subject is plus the 2–4 features that must survive the simplification (e.g. "a cat: pointed ear tips, rounded head, curled tail"). Never invent features that are not visible in the reference. 3. **Read the visual spec.** Call `read_skill_file(name='cookie-cutter-press', path='references/style-and-prompt.md')` before generating and follow its Style / Content / Layout / Must keep / Must avoid rules and its prompt template. Do not generate before reading it. 4. **Generate.** Call: `generate_image(prompt=<template from the spec, filled in>, input_urls=[<reference image>], model='i2i.hd.lite', orientation='square', name='<short title>', entity_type='cookie_cutter_images')` - `input_urls` carries the uploaded reference image (its asset_id, an `s3://` uri, or the asset of a previous result in this conversation). The reference image is the source of the cut outline. - `model='i2i.hd.lite'` — image-to-image, because the reference image must drive the outline. Never pass a physical model name. - `orientation='square'` — the layout is 1:1; leave `resolution` unset. - The prompt is always English and always built from the template in `references/style-and-prompt.md`; only the subject and signature-feature slots change, the fixed style and layout lines stay verbatim. 5. **Reply briefly.** One or two sentences describing the result in natural language. Do not paste URLs. ## Notes - Keep the subject recognizable and unchanged: same identity, same structure, same details the user emphasised. Only the material and presentation become "cookie-cutter press". - Add nothing the app does not have — no background scenery, extra characters, props, brand marks or text beyond the cutter and the dough. - Output is a single still image. Video and multi-frame requests are out of scope for this skill; say so instead of calling a video tool. - If `generate_image` returns `success=false`, report the failure plainly. Never claim success, invent an image, or fabricate a link. - Refuse violent / sexual / NSFW requests politely without calling the tool (`i2i.hd.lite` does not accept NSFW content). - When the user asks for a change, re-generate from the same reference with the adjusted instruction instead of describing a previous result as if it had been edited.