Update to ImagineArt

{
  "description": "Put a real garment from the user's own photo onto a model or person — virtual try-on, outfit swap, UGC try-on video. Use when the user supplies a photo of an actual garment or product they own. Not for generated wardrobe inside a styled shoot (use Fashion Shoot).",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 468
    },
    {
      "relative_path": "assets/ATTRIBUTION.md",
      "size_in_bytes": 945
    },
    {
      "relative_path": "assets/icon.svg",
      "size_in_bytes": 321
    },
    {
      "relative_path": "references/errors.md",
      "size_in_bytes": 2325
    },
    {
      "relative_path": "references/models.md",
      "size_in_bytes": 3809
    },
    {
      "relative_path": "references/qa.md",
      "size_in_bytes": 849
    }
  ],
  "name": "try-on",
  "skill_md_contents": "---\nname: try-on\ndescription: Put a real garment from the user's own photo onto a model or person — virtual try-on, outfit swap, UGC try-on video. Use when the user supplies a photo of an actual garment or product they own. Not for generated wardrobe inside a styled shoot (use Fashion Shoot).\n---\n\n# Try-On\n\nRead `references/models.md` before any video call.\n\n## Tokens\n\n`/try-on` `/outfit-swap` `/garment` `/wear-this`\n\nThe distinguishing signal: **the user has a photo of a real garment.** If the garment is\nbeing invented rather than supplied, this is Fashion Shoot instead.\n\n## When this fires\n\n- \"put this jacket on her\", \"how would this look on a model\"\n- \"try-on video\", \"outfit swap\"\n- A garment photo is attached and a person is mentioned\n\n## Truth and safety gate — clear this before generating\n\nIf any item fails, do not generate and do not route around it.\n\n- **Subject authorization.** Use a generated adult, or a real person the user is authorized\n  to depict. A supplied photo is not permission to impersonate its subject. Decline public\n  figures, minors, and deceptive identity use. If third-party consent is unclear, ask once.\n- **No synthetic testimonials.** A generated person is a model or demonstrator, never a real\n  customer. Do not invent ownership, results, or lived experience.\n- **Transparent framing.** Output is a brand concept or demo, not organic customer content.\n\n## Workflow\n\n1. **Upload the garment.** `request_image_upload` to render the upload widget. If the garment is shot on a\n   hanger, flat, or worn by someone else, call `extract_garment` first to isolate it.\n\n2. **Establish the person.** Either a model the user picks, or a photo of a person they\n   supply (collected the same way).\n\n3. **Compose.** `composite_shoot` with `image_urls` = **[garment, person]**. It requires\n   **two or more** images and runs a reference-aware reconciler so the person, the garment\n   and the setting stay consistent. Optional `camera_angle_id` from `list_camera_angles`\n   frames the shot; optional `background_select_id` from `select_fashion_background` sets\n   the scene.\n\n   For motion, generate the still first, then animate it with `generate_video`\n   (`image_url` = that one asset).\n\n4. Let the widget poll.\n\n## Get this right\n\n- **Never invent the garment.** The whole value is that it's *their* product. If the\n  uploaded photo is unusable, say so and ask for a cleaner one rather than generating an\n  approximation.\n- **Preserve garment detail** — print, logo, texture, colour. If the result drifts from the\n  source garment, regenerate rather than presenting it.\n- Default to `9:16` for try-on video; it is going to a social feed.\n- **Try-on is people.** Never open on `seedance-2.5` — it refuses human subjects, including\n  illustrated ones. Use `seedance-2.0` at 1080p, falling back to `ltx-2.3` at 6s/1080p.\n- **Never omit `resolution`** — it defaults to the model's lowest tier.\n\n## Conventions\n\n- `org_id` resolves itself; call the tool and retry with the id it returns if you don't\n  have one. Only call `select_organization` if resolution fails.\n- Never print ids.\n\n## Model refusals\n\nOn a safety or content refusal, stop and offer an alternative subject. Do not retry on\nanother model to bypass it. If the refusal reason is unclear, stop and clarify it.\nOnly switch models for a clearly established unsupported input capability, and tell the\nuser which model changed and why, including any duration or resolution change.\nSummarize the issue plainly instead of exposing the raw upstream error.\n\n## Tool availability\n\n`composite_shoot` requires **two or more** `image_urls` and a **paid organization**. There\nis no single-image try-on path — if you only have the garment, you must also establish the\nperson before compositing.\n\n## When something fails\n\nRead `references/errors.md`. Identify the kind of failure before responding: pending jobs\nare waited on, technical errors retried at most twice, unsupported capabilities stated\nplainly, safety refusals never routed around, and every material change disclosed.\n\n## QA before delivery\n\nRun `references/qa.md` before calling this finished. Inspect the actual output — never\nclaim quality you have not observed, and say plainly what you could not verify.\n"
}