Dreamina Canvas Discovery

This Skill teaches every other dreamina-canvas-* Skill how to ask the CLI what is currently supported. It exists so that no caller hard-codes model names, voice names, ratios, resolutions, durations, or batch limits.

Reference detail is in references/capability-discovery.md.

When to use

• Before the first generation in any workflow that takes a --model, --voice-name, --ratio, --resolution, --duration, or --count.
• Before any audio music generation (must confirm the active environment's audio model and that the caller has not cached a name from a different environment).
• When a generation fails with "model not found" or a --voice-name rejection: re-discover rather than retrying with a remembered value.