Media Content Workflows

Use this skill when the user asks to find, transform, analyze, summarize, or generate media-adjacent content: YouTube transcripts, GIFs, audio feature visualizations, lyrics, song prompts, or AI music generation.

Choose the path

• YouTube/video-to-text → fetch transcript first, then produce the requested output format (summary, thread, blog, notes, critique).
• GIF search/download → use Tenor-style search/download, verify file path/URL, and keep query terms safe and specific.
• Audio inspection → compute spectrograms/features (mel, chroma, MFCC) and report what was actually generated.
• Songwriting → craft lyric structure, prosody, rhyme, emotional arc, and genre tags before model prompting.
• AI music generation → prepare lyrics + tags + provider-specific constraints; verify returned files/URLs.

Workflow

1. Identify whether the deliverable is a file, a prompt, an analysis, or a transformed text artifact.
2. Retrieve or create the source material with tools; do not invent transcripts, downloads, or generated media.
3. Preserve reusable commands and output recipes in references/scripts.
4. Verify the artifact exists or the API returned a real handle.
5. Final response should include path/URL plus a concise description of content and any limitations.

Labeled playbooks

YouTube content

Fetch transcripts with the preserved script under references/packages/youtube-content/scripts/fetch_transcript.py, then format into summaries, blog posts, threads, timestamps, or study notes using the output-format reference.

GIF search

Use short descriptive search terms, inspect results before choosing, and download the actual GIF when the user needs a file.

Audio features

Use songsee-style CLI workflows for spectrogram and feature extraction. Mention sample rate, duration, and feature type when reporting.

Songwriting and AI music

Separate craft from generation: first write/refine lyrics and tags, then call the generation provider or prepare the exact prompt. HeartMuLa-specific details are preserved under references/packages/heartmula/; general songwriting guidance is under references/packages/songwriting-and-ai-music/.

Preserved source packages

Full prior skill packages are preserved under references/packages/: gif-search, heartmula, songsee, songwriting-and-ai-music, and youtube-content.