remix-graphic-ad-from-reference

Recreate a static graphic ad (Pinterest pin, IG/FB feed image, poster) from a reference image, swapping in a new brand's product and new copy while keeping the reference's layout, composition, and visual energy. ALWAYS generated with GPT Image 2 in edit-the-reference mode (fal-ai/gpt-image-1/edit-image, a billed FAL generation); the HTML/goose-graphics overlay is only an optional text-finishing step, never the generator. The static-graphics counterpart to the video remix-ad skill; this is what the app calls when a user picks a reference ad and wants it for their own product.

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

77%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A highly actionable, well-sequenced workflow with concrete commands and explicit validation gates. It loses points on conciseness (heavy repetition of the GPT-vs-HTML rule) and progressive disclosure (a largely monolithic body that keeps detailed reference material inline rather than splitting it out).

Suggestions

State the 'GPT Image 2 is always the generator; HTML overlay is only optional text finishing' rule once in Decision Rules and reference it elsewhere instead of restating it across Purpose, Phase 1, Phase 2A, and Failure Modes.

Move the detailed Brand grounding and Failure Modes sections into one-level-deep reference files (e.g. references/brand-grounding.md, references/failure-modes.md) and link to them from SKILL.md, keeping the main body a lean overview.

Trim the Spend Reference's credit math and redundant 'Phase 2A is NOT the generator' caveats now that the routing rule lives in one place.

Dimension	Reasoning	Score
Conciseness	Accurate and free of basic-concept padding, but the GPT-vs-HTML point ('ALWAYS GPT Image 2; HTML is only an optional finishing step') is repeated in Purpose, Phase 1, Phase 2A, Decision Rules, and several Failure Modes — 'mostly efficient but could be tightened', not the lean level-3.	2 / 3
Actionability	Gives fully executable, copy-paste-ready guidance: 'node <goose-graphics>/screenshot/screenshot.js --format <canvas> --input index.html --output render.png --font-delay 1500', 'fal-ai/gpt-image-1/edit-image', '--aspect_ratio 3:4 --quality high --resolution 2k', plus real scripts (scripts/cutout_product.py) and template (assets/overlay-template.html).	3 / 3
Workflow Clarity	Clear phased sequence (Phase 0 → 0.5 → 1 → 2A/2B → 3) with an explicit 3-check fidelity gate ('all must pass') and feedback loops (re-roll GPT or overlay text on failure), matching the level-3 anchor with validation checkpoints and error-recovery loops.	3 / 3
Progressive Disclosure	Sections are well-organized and the two bundle files (scripts/cutout_product.py, assets/overlay-template.html) are clearly signaled and verified real, but almost all detail (brand grounding, failure modes, QC) is inline in a ~230-line SKILL.md rather than split into one-level-deep reference files — 'content that should be separate is inline'.	2 / 3
	Total	10 / 12 Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, third-person description that covers concrete actions, natural trigger terms, both the what and the when, and a distinct niche clearly separated from its video counterpart. Its only weakness is implementation padding (FAL slug, 'billed FAL generation', goose-graphics) that a user would never say, though no rubric dimension is heavily penalized by it.

Dimension	Reasoning	Score
Specificity	Names multiple concrete actions — 'Recreate a static graphic ad... swapping in a new brand's product and new copy while keeping the reference's layout, composition, and visual energy' — matching the 'lists multiple specific concrete actions' anchor rather than the single-action level 2.	3 / 3
Completeness	Explicitly answers both what (recreate/swap a reference ad) and when ('this is what the app calls when a user picks a reference ad and wants it for their own product'), satisfying the explicit-trigger requirement that would otherwise cap this at 2.	3 / 3
Trigger Term Quality	Good coverage of natural user terms ('Pinterest pin', 'IG/FB feed image', 'poster', 'reference ad', 'wants it for their own product'); the level-2 anchor only has partial keywords, while this spans the common variations a user would actually say.	3 / 3
Distinctiveness Conflict Risk	Clear niche (remix a specific reference ad for a new brand) and explicitly contrasted with 'the video remix-ad skill', so it is unlikely to trigger for the wrong skill; level 2 would still risk overlap with similar skills.	3 / 3
	Total	12 / 12 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: gooseworks-ai/goose-skills
Path: skills/ads/composites/remix-graphic-ad-from-reference/SKILL.md
Commit: 8866b2a

Reviewed: 1 day ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.