Media Generation Capability Resolver (design/MediaGenerateDialog.logic.ts)

Infers audio type from prompt prefixes and selects appropriate image generation or editing models based on reference image count.

// Source: https://github.com/shiwenwen/hope-agent | License: MIT | Copyright 2026 shiwenwen
// Provided via justclaw.it.com; keep this notice with the code.
import type {
  ImageEditCaps,
  ImageModelCaps,
  MediaAudioKind,
  MediaCandidateOverview,
} from "@/components/settings/media-gen/types"

export function inferAudioKindFromPrompt(prompt: string): MediaAudioKind | null {
  const lead = prompt.trimStart().toLowerCase()
  return lead.startsWith("[music]") ? "music" : lead.startsWith("[sfx]") ? "sfx" : null
}

export function resolveImageRequestCapability(
  candidates: MediaCandidateOverview[],
  referenceImageCount: number,
): {
  candidate: MediaCandidateOverview | null
  caps: ImageModelCaps | ImageEditCaps | null
} {
  if (referenceImageCount <= 0) {
    const candidate = candidat

... (truncated -- full source via MCP)