Chibird AI Short Drama Production

Problem It Solves

Short-drama, poster, and cover generation requests are often split between a web console and API docs: submission, polling, history lookup, failure handling, and point-balance messages all need manual coordination. This skill brings Chibird AI / Chibird Mancang image and video generation into the conversation layer, which is useful when you need to pass UTF-8 Chinese prompts to the backend, track task status, and return result links without re-implementing the workflow.

How It Works

• Image generation: submit redbird_image_cover, poll redbird_image_poll, and query history with redbird_image_list.
• Video generation: submit redbird_video_upload, poll redbird_video_poll, and query history with redbird_video_list.
• Parameters: use skill.yaml defaults when unspecified, and let user-provided values override them. Prompts should be passed as UTF-8 Chinese text.
• Models: images can use N-2, N-Max, G-2, or M-5; videos can use All-in-One Video 2.0 or All-in-One Video Mini.
• Polling: check at 8-second intervals, up to 30 attempts for images and 60 for videos. Stop when status is 1 and the link is valid, or read error details when it is 2.
• Constraints: submit cover/upload only once per user request, and do not auto-resubmit after failure or timeout unless the user explicitly asks to regenerate.

Boundaries

The skill focuses on API orchestration and status handling rather than creative scriptwriting. Video generation is billed by duration, results can occasionally fail or time out, and history should be checked when uncertain. Insufficient points require handling on the Chibird Mancang official site. User-facing wording should use Chibird or Chibird Mancang instead of translating the brand.