菜单
Generate one prompt-based image edit from explicit input images.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | image-blast-image-edit |
| description | Generate one prompt-based image edit from explicit input images. |
Use this skill for source cleanup, object references, clean plates, masks, or standalone image edits.
Require at least one input image and one edit prompt.
Inspect the output directory with ls -a.
Use explicit --role and --output-slug values when the caller provides
semantics such as plate, object-mask, or image-edit.
Run:
node .claude/scripts/image-edit/generate-edit.mjs \
--image "<input image path>" \
--prompt "<edit prompt>" \
--output-dir "<output directory>" \
--role "<role>" \
--output-slug "<output slug>"
Optional provider override: --provider nano-banana|gpt-image-2.
Repair missing local image files from request metadata when needed:
node .claude/scripts/project/ensure-local-assets.mjs --from "<request-json-path>"
Report input images, output image, request metadata, role, and prompt used.
Generate one atomic 3D object from an approved object.json or direct source image.
Generate one clean plate image for a world by removing confirmed foreground objects or user-specified content.
Create, inspect, and manage an image-blaster project envelope under worlds/<slug>. Use before other image-blast skills or whenever the user asks about project state.
Generate world ambience, object impact sounds, or custom sound effects through the FAL ElevenLabs SFX endpoint.
Analyze source images, write image.json, present object candidates, and create object.json files after user confirmation.
Discover and run arbitrary FAL operations that do not fit the narrower image-blast skills.