flora-mcp-skills: Skill for Claude Code

flora-mockup-deck

Attribution. Pass skill: "flora-mockup-deck" on every FLORA call you make while running this skill — execute included — along with a skill_run_id you invent once when the run starts and reuse for the rest of it. Both are reporting only: they change nothing about the call or its result.

What it is

Not an image generator. A placement engine. By the time the placements fire, the creative is finished and frozen — the only job is putting it into the world convincingly, four ways, without altering it.

Usually the creative arrives finished. It can also be written: a line of text buys one generation, and that output becomes the master creative, from which point it is handled exactly like a supplied file. See "Two ways in".

The thing that makes a mockup fail is never the photograph. It is the artwork drifting: recoloured, recropped, re-lettered, a word dropped. So the whole skill is built around one invariant.

Two ways in

ARTWORK   a url or an attached file      -> 4 generations
WRITTEN   a line of text, no artwork     -> 5 generations (master + 4 placements)

/flora-mockup-deck followed by a description and nothing attached is the WRITTEN path: create a campaign deck of a flower, a poster for a late-night ramen bar, out-of-home for a running shoe launch. Nothing else about the skill changes. The text buys exactly one image, that image becomes the master creative, and the law closes over it.

Making the master

One flora_generate, text-to-image, same model family as the placements:

model   "t2i-gpt-image-2-t2i"
params  { aspect_ratio, resolution: "4k", quality: "high" }

Prompt it as a printed poster, not as a photograph of a thing. The text names a subject, and a subject is not an ad — "a flower" un-elaborated returns a stock botanical photograph, which then gets pasted onto a gable end and reads as a picture of a flower on a wall rather than as a campaign. Build the master prompt as: