| id | skill-raster-ui-asset-design |
| type | agent-skill |
| status | reviewed |
| created | 2026-06-21T00:00:00.000Z |
| updated | 2026-06-21T00:00:00.000Z |
| name | raster-ui-asset-design |
| description | Design and integrate generated raster images for child-agent UIs only when the asset supports a concrete workflow state, preview, comparison, consequence or operator control. |
| version | 0.1.0 |
| topics | ["agent-skills","raster-ui-assets","image-generation","child-agent-ui"] |
| tools | ["Pritha","Codex","imagegen","OpenAI image_generation"] |
| sources | ["04_standards/raster-ui-assets-for-child-agents.md","07_workflows/raster-ui-asset-generation.md","03_reviews/2026-06-21-raster-image-generation-ui-source-batch-review.md"] |
| related | {"workflows":["07_workflows/agent-skill-pack-selection.md","07_workflows/raster-ui-asset-generation.md"],"standards":["04_standards/raster-ui-assets-for-child-agents.md","04_standards/agent-interface-experience.md"]} |
| source | pritha-memory |
| source_paths | ["04_standards/raster-ui-assets-for-child-agents.md","07_workflows/raster-ui-asset-generation.md","03_reviews/2026-06-21-raster-image-generation-ui-source-batch-review.md"] |
| review_status | reviewed |
| trust_level | local-reviewed |
| requires_toolsets | ["filesystem","markdown","image-generation"] |
| risk_level | medium |
| tags | ["raster-assets","ui","imagegen","accessibility"] |
Raster UI Asset Design
When to Use
Use when Pritha or a child agent is designing a non-chat UI and a raster image
may help users inspect workflow state, compare visual candidates, preview
generated artifacts, understand consequences, or review media/product/lesson
content.
Procedure
- Start from the child-agent workflow, not from visual style. Record user
journey, visibility needs, control needs, state model, side-effect policy,
rendering boundary, framework/stack, fallback, mobile/accessibility, privacy
prompts and readiness check.
- Reject raster if DOM/CSS/native UI or SVG would satisfy the job more clearly.
- Write a raster asset spec before generation: target component, state shown,
user-visible purpose, controls around the image, output sizes/formats,
reference image roles, alt/fallback and privacy boundary.
- Use prompt-plus-controls in the dialogue: style chips, aspect ratio, generate
or edit mode, preview thumbnails and accept/reject/regenerate/edit actions.
- Use Codex
imagegen for normal project-bound assets. Use API/CLI fallback
only when selected by the contract and current docs have been checked.
- Save accepted assets in the child-agent workspace and wire them into real UI
components. Keep text, controls, warnings and status outside the image.
- Verify mobile crop/readability, responsive variants, file size, format
fallback, alt/fallback correctness, privacy retention and user controls.
- Record final path, prompt/spec, variants, tool path and verification result in
the scaffold report or UI notes.
Pitfalls
- Do not generate decorative dashboards just because image generation is
available.
- Do not hide action buttons, warnings, logs, task state or approval text inside
pixels.
- Do not send private screenshots or reference images to hosted generation
without a contract-selected privacy boundary.
- Do not leave project-referenced generated images outside the workspace.
- Do not treat AVIF/WebP, transparency or image-generation model options as
timeless; recheck current docs before implementation.
Verification
- The asset has a workflow purpose and target component.
- Real UI owns controls, text, status and side effects.
- Final assets live in the workspace with stable paths.
- Responsive, accessibility, file size and privacy checks are recorded.
- The user can accept, reject, edit or regenerate generated candidates when the
asset is part of an interactive design flow.
