---

name: screen-reference description: Generate, finalize, and validate a non-pixel-art full-screen visual reference for scene direction and evaluation, without producing a runtime asset.

Screen Reference Asset

Use this Skill for a full-screen visual anchor for scene direction and visual evaluation. It is reference-only: it does not create a runtime Godot resource, has no godot_artifact, and must not enter worker runtime handoff. Do not compile it to Texture2D, AtlasTexture, or any other native runtime resource.

Do not request or produce pixel art. Do not use nearest-neighbor scaling to imitate pixel art. Raw visual material may come only from the selected image provider or an explicitly supplied user source/reference. Do not create art, placeholder images, colour blocks, SVGs, canvases, Godot drawings, or image assets with ad-hoc scripts.

Invocation

Accept one request matching the shared Asset Skill request schema in .godotmaker/asset-runtime/schema/asset-skill-request.schema.json. Require asset_type to be screen-reference. references is optional. When it is present and non-empty, every { role, path } is a required visual input:

1. resolve the path inside the project and require a readable image before calling a provider;
2. preserve its canonical, style, or screen role in the prompt and source-generation provenance;
3. attach the actual image bytes to the selected provider request. A path mentioned only in text is not an attachment;
4. STOP before writing a raw source or final output when an input is unreadable or the selected provider cannot attach it.

Use the configured provider named by the request or production brief. native, codex, gemini, openai, and wan are binding selections: execute that documented path or STOP. Never silently switch providers. For codex, non-empty references require the image_gen call's referenced_image_paths argument.

This Skill can be invoked directly or by an orchestrator with the same contract. Do not read or write ASSETS.md, tags, stage state, generated indexes, or worker dispatch state. The /gm-asset manager later registers its validated generic result directly in the matching catalog row.

Production contract

spec must declare the target size (WIDTHxHEIGHT) and aspect_ratio (WIDTH:HEIGHT). Build a prompt that names the game and screen purpose, camera/viewpoint, visible gameplay objects, approximate layout, HUD or UI safe regions, style language, target aspect/orientation, and each supplied reference role. Do not add labels, callouts, debug overlays, or unrequested objects.

Treat object positions and sizes in the provider image as approximate visual direction. The raw provider source is not required to satisfy the final canvas or object pixel dimensions. Once the provider returns a readable image, claim it to the deterministic raw-source path before applying final canvas checks. Pixel-exact runtime sprite dimensions remain the responsibility of runtime Asset Skills such as compact-prop-pack and fx-bundle, not this reference.

For every accepted image, retain these deterministic paths:

1. prompt: .godotmaker/asset-generation/prompts/<asset_id>.txt;
2. raw provider source: .godotmaker/asset-generation/sources/<asset_id>_source.png;
3. source/provider report: .godotmaker/asset-generation/reports/<asset_id>_source.json;
4. final reference: references/<asset_id>.png;
5. finalize report: .godotmaker/asset-generation/reports/<asset_id>_finalize.json.

For openai, gemini, and wan, write a source-generation spec using reference_inputs for role-preserving references, then run:

python tools/asset_source_generate.py --spec <spec.json>

The successful source report must identify the selected provider and model, actual reference attachment count, each reference role/path/hash, raw source, and prompt path. For native and codex, use their provider documents and write an equivalent JSON report before finalization. It must contain ok: true, asset_id, raw_source, raw_source_sha256, reference_attachment_count, and a generation object with tool: "image_gen". When references are supplied, that object must also record reference_attachment_argument: "referenced_image_paths" and the matching attachment count, and each reference_inputs item must preserve role, path, sha256, and attached: true. A missing report, provider failure, or absent attachment evidence is a STOP.

Finalize only the provider/user source with the existing controlled tool:

python tools/asset_image_finalize.py \
  --source .godotmaker/asset-generation/sources/<asset_id>_source.png \
  --out references/<asset_id>.png \
  --label <asset_id> \
  --require-aspect <WIDTH:HEIGHT> \
  --resize <WIDTHxHEIGHT> \
  --fit cover \
  > .godotmaker/asset-generation/reports/<asset_id>_finalize.json

The explicit cover fit proportionally scales and center-crops the raw image to fill the final canvas without transparent padding. After finalization, require the report dimensions to equal spec.size and validate the finalized reference. If aspect validation, finalization, or finalized-output validation fails, STOP. Keep the captured finalize report as result evidence; the reference result has no runtime artifact and is registered as source_ready by the manager.

Result

Return the shared generic result with one reference output, its deterministic raw source, no godot_type, and no runtime outputs. Run standalone_validation.compile_and_validate() before returning it; it replaces any caller-supplied validation claim and proves the stable final reference and raw source are readable PNG files. It begins from the shared result schema and checker, then proves the result.

{
  "asset_type": "screen-reference",
  "outputs": [{
    "role": "reference",
    "name": "<asset_id>",
    "path": "references/<asset_id>.png"
  }],
  "sources": [{
    "path": ".godotmaker/asset-generation/sources/<asset_id>_source.png",
    "layout": "single"
  }],
  "previews": [],
  "validation": {
    "passed": true,
    "levels": {"L0": true, "L1": true}
  }
}

L2-L4 are not applicable because a screen reference has no Godot artifact and is never handed to a worker. Preserve the provider and finalize reports so the manager and private Eval can audit the production gates separately; L5 is not applicable and L6 is visual/semantic review.