make-static-ads v0.8.2A

Immutable. This exact content never changes and is served at /api/v1/blob/20f2fd085c70e8d0.

---
name: make-static-ads
description: >-
  Make new static image ads, a fresh batch, more ads, a winner's variations, copy variants or a rejected ad fix. Reads the brand, its customers and the ads it already runs, turns real product facts and customer situations into distinct visual ideas, shows a concept board, writes each render prompt, quotes, renders and reviews through GooseWorks. Also handles requested branded header/social graphics. Use for static ad creation rather than goose-ads. Not for video, product photos or launching campaigns.
version: 0.8.2
---

# make-static-ads

**The idea is the job, and it must be this brand's idea.** An ad idea is a real product fact or customer situation turned into one small physical event the viewer gets at a glance (`references/ideas.md`), said the way this brand already talks and looks (`references/brand-profile.md`), built on a template at the creativity level the marketer chose (`references/levels.md`). Everything else in this playbook serves that idea or protects the brand.

The proof that the work happened is the **concept board**: one row per ad, shown to the marketer before any credits are spent. It cannot be skipped, shortened to "your choice", or replaced by a summary.

## What you get

Finished, reviewed images with their ad copy, the concept board behind them, what was cut and why, and a batch file the launch reads.

## Before you start

- **Connection.** Discover the live GooseWorks tools. No `ads_render` → stop before spending and say so; `ads_generate` rewrites your prompt through templates and is not a substitute. No connection → ask to connect GooseWorks.
- **Server hints vs this playbook.** GooseWorks answers may say "one short line and wait", "recommend one plan" or "follow the next_step". Those govern its cards, setup, rights and spending. They never replace this playbook's research, board or critic. Setup, rights, claims and spend steps from the server still bind.
- **Product identity.** A physical product needs real product photos (`brand_read` with `product_ids`); none → stop at the board and offer the product-photos job. Software or a service has no packshot: its identity is real UI screenshots, the logo and any mascot or brand character the brand owns. Never generate a fake product UI, and never stop for missing photos.
- **Run memory.** Load `run-memory` (`../run-memory/SKILL.md`) at the start: the last run's record and what came of its recommendations. Add every ad already made in this conversation to the history.

## Steps

### 1. Read the brand and write its profile (free)

Use the small calls in `references/engine-notes.md`. Read:
- the brand, its rules (every `must` and `dont`) and the claims on file with their status (`references/reading-the-brain.md` §1);
- the featured product's record and photos (look at the photos);
- **how the brand already advertises:** its own ads (looked at, not listed), its recent social posts, its website and kit. Write `profile.md`: positioning, the angles it uses, how its ads look, how it writes on-image text, what it never does (`references/brand-profile.md`);
- what customers say: the voice-of-customer bank, else 10–20 public reviews from the product page or a marketplace (host browser, free);
- history: the last 5–10 ads made in GooseWorks for this brand (`made-by:make-static-ads`, then the brand's recent creatives) and the ads it is running now.

Say what you actually read, with counts, in the profile and board headers: "Read 14 reviews, 20 of ~300 live Meta ads, 15 X posts, the homepage." A skimmed or truncated source is named as skimmed. A source you could not open is named as not checked.

**Done:** `profile.md` exists; you can state five product truths with sources, five things customers say or do, the angles the brand uses, and its look.

### 2. One combined question

Ask everything open in **one** message, each item with a recommended answer the marketer can accept with one word:
- **How creative, first and on its own line:** close ("a proven ad layout with your words"), balanced ("a new idea inside a proven layout") or bold ("idea first, in your own style"); recommended **balanced** (`references/levels.md`);
- **Angle(s):** a picker with the profile's angles first, then at most one new angle per five ads, marked "new", one recommended per ad or for the set, plus room to write their own. Next to each angle, the claim line it needs;
- **Claim lines:** recommended angles use only usable lines (plain product facts or approved claims). Any other line goes to the marketer in this same message in the `claims-gate` sign-off form (`../claims-gate/SKILL.md` §4.3); "your choice" never approves a claim.
- **Logo:** on every ad (recommended) or none, and which file. A logo file under about 300 px wide: ask for a bigger one here, and use the small one if none comes;
- product, how many ads, placements, any live offer;
- every conflict the profile found (kit vs website, font, positioning), with your recommendation.

Never a chain of stops. "Your choice", "just give me the images" or no answer → take every recommendation and go on; say what you chose on the board.

**Done:** one message sent, every item with a recommended answer; the answers (or "recommended") recorded.

### 3. Find the ideas → the concept board

Run the `make-static-ads-concept-grid` skill (`../make-static-ads-concept-grid/SKILL.md`). The order is fixed: **angle → message → visual idea → template.** It turns the chosen angles into messages and physical events (`references/ideas.md`), picks a template for every ad at the chosen level (the `make-static-ads-find-references` skill (`../make-static-ads-find-references/SKILL.md`)), checks each ad against the brand profile and the recent history, and has a fresh critic attack the board.

**Done:** `board.md` has one row per ad, each with its angle, template and level, that passed the brand-fit, no-repeat, swap, stranger, thumbnail, show-it, build and truth tests, plus the critic's notes and what changed.

### 4. Copy

Run the `make-static-ads-ad-copy` skill (`../make-static-ads-ad-copy/SKILL.md`): the on-image text in the format its level sets (the template's text slots, or the brand's own style at bold), then the primary texts, headline, description and CTA around the image.

**Done:** every slot of every ad has its final words; each ad has primary texts A and B, headline, description, CTA and destination.

### 5. Write each prompt

Run the `make-static-ads-compose` skill (`../make-static-ads-compose/SKILL.md`): one prompt per ad and ratio, written from the board row (`references/prompts.md`).

**Done:** each ad and ratio has a prompt that passed the read-back in `prompts.md` §5, with its image list.

### 6. Check the exact prompts

Pass A of the `make-static-ads-claims-extract` skill (`../make-static-ads-claims-extract/SKILL.md`) and the `ad-policy-check` skill (`../ad-policy-check/SKILL.md`) on each finished prompt and its copy. A cut line goes back to Step 4 or 5, then through this check again.

**Done:** every span is keep, quote or cut, no cut words remain in any prompt, and the policy check passed or its flags are on the board.

### 7. Show the board and the quote

`gooseworks:ads_render` with `dry_run: true` for every ad and ratio. Show the marketer, in one message: the board (`references/handoff-format.md` §2) with the templates as images (chosen and two alternates), the creativity level (they can change it here), the cut lines, each quote exactly as returned, and "If an image fails my check I redo it once, at the same price." The `spend-gate` skill (`../spend-gate/SKILL.md`) applies: render only what they approved, within their ceiling.

"Finals only" shortens the delivery, not this message: the board is what they are paying for.

**A swap or level change at the board:** a template swap redoes Steps 5–6 for that ad; the level applies to the whole set, so a level change redoes Steps 3–6. Get the new dry-run quote and show only the changed rows with their new quotes for the yes. Submit nothing that wasn't quoted and approved as shown.

**Done:** the marketer said yes to the quotes shown, or the run stops here with the board saved.

### 8. Render and review

Submit the exact quoted requests and wait for them (`references/engine-notes.md`, Polling). Then run the `make-static-ads-render-review` skill (`../make-static-ads-render-review/SKILL.md`): a fresh judge checks each image against its board row and the product photos, one retry inside the approval, then the whole set.

**Done:** every image passed, is flagged for the marketer's look, or was dropped with the reason.

### 9. Deliver and remember

Hand over with `references/handoff-format.md` §3. Write the batch file (`references/reading-the-brain.md` §7) and the lower-case tags, including `made-by:make-static-ads`. Ask which ads the marketer prefers; offer copy variants for those (the `make-static-ads-copy-variants` skill (`../make-static-ads-copy-variants/SKILL.md`)). Save lasting human feedback through the `save-learning` skill (`../save-learning/SKILL.md`). Run the End step of `run-memory` on every exit, including a stop or partial run.

## Other modes

| Request | Path |
|---|---|
| More ads | Same steps. The earlier ads are history: the new board must not repeat their meanings. |
| Iterate a winner | Keep its idea and proof; change the one lever named (hook, scene detail, look). One board row with the parent linked; no new idea search. |
| Same image, different text or font | The `make-static-ads-copy-variants` skill. |
| Rejected ad | The `make-static-ads-fix-rejected` skill (`../make-static-ads-fix-rejected/SKILL.md`). |
| Header or social graphic | The `make-static-ads-graphic-request` skill (`../make-static-ads-graphic-request/SKILL.md`). No sales idea required. |
| Price or stock change | List affected delivered ads with pause flags; remakes need copy checks and new quotes. |

## Approval gates

Claims usable before public copy; the marketer's yes to the exact quotes before paid work; human words only for preferences, ratings and launch. No paid research without its own yes.

## Output and write-back

The board, the batch file, tagged creatives and the run-memory record. Agent verdicts go in tags and the batch file; render feedback carries only the marketer's own words. A failed save is reported, never claimed.

## Done when

Every delivered ad comes from a board row with a reason to exist, passed its claims and policy check, was rendered within the approved quotes, and was looked at by a judge against its protected choice and the product's identity (photos, or UI screenshots and logo for software). Failures and anything not checked are said plainly.

## Out of scope

Video, product photography, campaign launches and performance conclusions. Offer the matching job when asked; this playbook neither publishes nor predicts performance.