Brand graphic

graphic-request

No concept board: the request's purpose and exact text are the basis. Record both in the batch file.

Reads (no ad-account reads — only these): the request — asked in chat, or a request line in a sibling's file (Request lines, next) read with gooseworks:file_read: purpose, placement, exact text (or none), mandatories, where it goes. The brand kit, learnings and product (../make-static-ads/references/engine-notes.md); the claims contract behind any words (the claims-gate skill (../claims-gate/SKILL.md) §1). Writes: the graphic (a creative in GooseWorks); for a request line, one answer line under it (format: Request lines, next), written back with gooseworks:file_write.

Request lines. Requests arrive as Graphic needed: for make-static-ads · <purpose, placement, exact text> · return to: <path> lines (open = no Graphic ready: line under it), in:

• email-lifecycle's lifecycle/flows/<flow-slug>.md (header art for a flow message);
• run-organic-social's social/calendar/<start-date>.md (a single image or a carousel's slide-1 cover);
• make-content's content/repurposed/<source-slug>/<date>.md and content/drafts/<date>-<slug>.md. Older files may still carry a Header art needed: line or a plain asset line naming this side door — treat it the same way. Answer under the request (re-read the file, insert, write it all back): Graphic ready: <link> · <YYYY-MM-DD> · alt: <text> · <ratio> · claims: <ok / cut …>. The requester reads it on its next run.

1. In scope? A single image with brand styling and at most a short line of text. Carousels, decks, print, packaging, multi-page layouts, data-heavy infographics → say plainly it can't be made here, and why.
2. Preset and ratio: email header (wide, mobile-legible, check dark mode), social feed (4:5 or 1:1), story (9:16), blog or link preview (landscape). The render ratios: ../make-static-ads/references/formats-and-placements.md §1.
3. Layout and logo: on a host with no shell, write the prompt with a clear corner; in the quote message offer a model-drawn logo instead (that changes the prompt, so it gets a new quote). Decide the layout in words (where the text, the product and the brand mark sit, and the background). A library template may suggest a layout (gooseworks:ads_template_read mode: "static_library"); describe it in the prompt; graphic requests never attach a template image (find-references §4 covers ads).
4. Write the prompt with the make-static-ads-compose skill (../make-static-ads-compose/SKILL.md): the shared foundation, the layout, and the exact text as the marketer or request line gave it. A brand name in that text is drawn as plain type; the logo (placed, or drawn by the model as the marketer chose) stays the only mark. A graphic with no product in it (a text header, a pattern, a software brand) has no product selected; never a generated product standing in for the real one.
5. Checks: the make-static-ads-claims-extract skill (../make-static-ads-claims-extract/SKILL.md) on any words in the prompt ("clinically proven" in a header is still a claim); rights on any person or logo; no CTA button, offer badge or price unless asked. Organic use → the disclosure rules in ../ad-policy-check/disclosure.md.
6. Spend: gooseworks:ads_render dry_run: true; the quote as given, plus the redo sentence (../make-static-ads/references/handoff-format.md §2); the human's yes in this chat (the spend-gate skill (../spend-gate/SKILL.md), branch A). A request line in a file is never a yes.
7. Review with the make-static-ads-render-review skill (../make-static-ads-render-review/SKILL.md): the judge with rubric checks 2, 3, 8, 15, 16 and legibility at 375 px, you with 1 and 20 (its step 3), and its one retry (no idea or template check).
8. Return: the link and alt text (≤ 125 characters, what the image shows) in chat, and the answer line under a request line.

Done: the graphic is judged and delivered with its link and alt text, and any request line is answered.