{"id":99227,"plugin_id":"plugin_asdk_app_6aee1baa4d8c8191aad5148cb8dd31fe","kind":"skill","collection_source":"plugin_package","comparison_source":null,"observed_at":"2026-11-16T18:17:05.153Z","digest":"8930fe81bc07aca525fe6edb412a32a1f3284500254908b8e4a668b6b7c261d3","against":null,"payload":{"description":"Create coordinated narration, music, and sound effects with idea.ai from a creative brief, with deliberate voice, timing, and cost choices.","included_files":[],"name":"idea-ai-audio-production","skill_md_contents":"---\nname: idea-ai-audio-production\ndescription: \"Create coordinated narration, music, and sound effects with idea.ai from a creative brief, with deliberate voice, timing, and cost choices.\"\n---\n\n# Audio production pack\n\nRead [the shared MCP workflow](../../references/mcp-workflow.md) before calling tools. It defines discovery, file handling, cost authorization, waiting, and delivery for this skill.\n\n## MCP operation\n\nUse `create_audio` for each narration, music, effects, or cleanup asset. For uploaded sources, use measured upload metadata in `quote_cost` before approving a media-to-audio operation.\n\n## Turn the brief into an audio plan\n\nIdentify the requested assets: narration, a music bed, sound effects, or a subset. Establish language, duration, mood, pace, instrumentation, loop requirements, and intended use. Keep an explicit asset list with one purpose and duration target per item.\n\n## Inspect the relevant audio models\n\nDiscovery examples are ElevenLabs v3 (`eleven_v3`), ElevenLabs Music (`eleven_music_v1`), and ElevenLabs Sound Effects (`eleven_sound_effects_v1`). Inspect actual models before choosing options. Music exposes `duration_seconds` and `force_instrumental`; sound effects exposes `duration_seconds` and `loop`.\n\nFor speech, obtain the model's voice catalogue and real voice IDs through `list_voices`. Use facets to resolve missing preferences and offer a small selection instead of silently picking from hundreds of voices. Keep script wording distinct from voice direction.\n\nQuote each asset using its actual prompt or structured `inputs` text/voice list and complete options. For structured dialogue, retain `inputs` when merging resolvedOptions; a separate prompt is optional. For prompt-priced speech, include the actual script rather than letting an omitted prompt select a maximum-length quote. Explain any reservation and settlement conditions. A revised script can change the price.\n\n## Review and deliver the assets\n\nCheck spoken wording, pronunciation, pacing, requested duration, clipping, noise, musical direction, and loop seams where playback is available. For existing noisy media, VEED Clean Audio (`fal_veed_clean_audio_v1`) exposes normalization, strength, and loudness controls; treat cleanup as an additional scoped action.\n\nDeliver separate, clearly named audio files with their purpose and timing notes. A general multitrack mixer is not exposed by the inspected tools. Do not claim that independently generated music, narration, and effects have been mixed or synchronized into a finished soundtrack unless an available additional tool actually performs and verifies that work.\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}