generate-image

Constraints

• The model's final user-facing response should match the user's input language, default English.
• Must request user confirmation before triggering any paid/high-cost image generation call.
• After confirmed, must call CreatOK Open Skills proxy and wait until completion.
• Avoid technical wording in the user-facing reply unless the user explicitly needs details for debugging.

Model Selection Rules

• Seedream 5.0 Lite

  • actual model id: seedream-5.0-lite
  • faster and lighter, good for quick iteration
  • resolutions: 2K, 4K only

• Nano Banana Pro

  • actual model id: nano-banana-pro
  • highest quality, best for photorealistic portraits and product shots
  • resolutions: 1K, 2K, 4K

• Nano Banana 2

  • actual model id: nano-banana-2
  • latest Nano Banana, best overall quality
  • resolutions: 1K, 2K, 4K

The model should recommend a model before generation based on the use case:

• portraits / photorealistic people → nano-banana-2
• product shots / e-commerce → nano-banana-pro
• general illustration / concept art → nano-banana-2
• quick preview / iteration → seedream-5.0-lite
• user explicitly wants 1K → nano-banana-pro or nano-banana-2 (Seedream does not support 1K)

Inputs to clarify (ask if missing)

• ask only for what is necessary to generate a good image
• if resolution or aspect ratio is not specified, use sensible defaults (2K, square)
• if the prompt is vague, offer to refine it before confirming generation
• reference images are optional — ask only if the user implies style transfer or subject reference

Workflow

1. Confirmation gate (mandatory)

• Summarize:
  • model
  • resolution
  • number of images (n)
  • aspect ratio if specified
  • estimated cost/credits if available
• Ask for a simple confirmation before submitting.
• Do not submit the generation task until user says yes.

2. Submit image generation

• Call CreatOK: POST /api/open/skills/image-generation

3. Poll status until completion

• Call CreatOK: GET /api/open/skills/tasks/status?task_id=...&task_type=image_generation

4. Persist artifacts + respond

• Write:
  • outputs/result.json with task_id/status/images/raw
  • outputs/result.md
• Persist the task_id immediately after submission so the user can recover later.
• Return the final image URLs.

Existing Task Recovery

• If the user already has a task_id, continue from that task instead of starting a new one.
• In recovery mode, do not ask the user to restate the prompt if the task id is already available.
• The model can either check status once or keep polling if the user wants to wait.
• If the task succeeded, return the final image URLs.
• If the task is still queued or running, explain clearly and offer to keep checking.
• If the task failed, explain the failure and suggest next steps.

Artifacts

All artifacts under generate-image/.artifacts/<run_id>/....

Thin Client Boundary

• This skill submits generation jobs, polls status, and persists fixed-format outputs.
• The model should not make the user restate their idea if the direction is already clear from the conversation.