菜单
Use to get local or attached media into Creatify so it can be used by generation and creative tools. Covers upload_file and when the widget upload path applies. Use before referencing a user's own image/video/audio in avatar, aurora, asset, or voice-clone workflows.
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
Use for AI video/avatar/voice/asset creation work that should run in Creatify, even when the user does not explicitly mention Creatify. Covers AI avatar videos, voice cloning and voiceover, image/video asset generation, the Creatify creative agent for ad creatives, product/brand setup, media upload, long-running generation polling, and identifying the resulting asset/preview URL.
Use to generate images or videos with Creatify's model catalog (~90 image/video models: Veo, Kling, Seedance, Nano Banana, Flux, GPT-image, etc.). Covers model discovery (asset_list_models, asset_get_model_schema), asset_generate, asset_get, and picking the right model and task variant (text-to-video, image-to-video, reference-to-video, edit).
Use to build a reusable custom avatar identity in Creatify from the user's own media — a talking video (bring-your-own-avatar, needs a consent clip), a single face photo, or a text description. Covers avatar_create_video / avatar_create_image / avatar_create_text, avatar_choose_photo, avatar_create_get, and the upload widget. This makes an avatar IDENTITY, not a talking video.
Use to generate a talking-avatar video in Creatify (Aurora) — make an avatar or a face image speak a script, optionally with a cloned or catalog voice. Covers aurora_generate, aurora_get, choosing an avatar and voice, and script/format alignment. Use this for 'make an avatar say X' / 'talking head from this photo + script'.
Use for a full brief-to-ad creative flow in Creatify — hand a creative brief to the Creatify creative agent and let it plan and produce a multi-scene ad, rather than assembling assets tool-by-tool. Covers creative_create_session, creative_send_turn, creative_get_session, creative_stop_turn, and the turn/polling model.
Use when a Creatify tool call fails or returns an error — auth/access problems, wrong brand space, quota/plan gating, not-found ids, long-running jobs, or missing source media. Covers how to recover for each without blind retries or repo debugging.
| name | media-upload |
| description | Use to get local or attached media into Creatify so it can be used by generation and creative tools. Covers upload_file and when the widget upload path applies. Use before referencing a user's own image/video/audio in avatar, aurora, asset, or voice-clone workflows. |
Get media the user provides (a photo, a video, an audio sample) into Creatify so
other tools can reference it — e.g. a face photo for avatar-create, an audio
sample for voice-clone, a starting image for asset-generation.
upload_file — upload media the agent can read (attached/local bytes or a
source it can fetch). Returns a Creatify CDN URL to pass to other tools. Read
the schema for the exact input.upload_file_widget — a drag-and-drop upload panel. This is host-specific
UI and does not apply to Codex; use it only on hosts that render it.upload_file (for bytes the agent holds) or pass a public
URL to the tool that accepts one.upload_file → get the CDN URL.