music-discovery

Formulate semantic music search queries, expand genres/moods, and discover music using alx search CLI.

SKILL.md

GitHub

Music Discovery Skill

Overview

This skill teaches AI agents how to act as an intelligent music curator. Instead of performing plain literal searches, the agent translates high-level user requests, moods, scenarios, or abstract genres into optimized search queries, using alx search to discover and play matching music.


Agent Behavior & Search Expansion

When a user asks for music based on a mood, activity, or vague description, the agent MUST expand it into high-affinity keywords (Artists, Genres, Album keywords, or precise titles) rather than querying the exact phrase.

Query Expansion Table

User Request Key Mood / Genre Search Term Recommendation Command Example
"Play something to help me concentrate" Lofi / Ambient / Minimalist "lofi study", "ambient focus", "chillhop" alx search "lofi study" --source wy
"I need high energy gym music" Synthwave / Hardstyle / Phonk "phonk workout", "synthwave running", "cyberpunk" alx search "phonk workout"
"Show me classic Chinese rock" 90s Chinese Rock / Folk "崔健", "唐朝乐队", "魔岩三杰" alx search "崔健"
"Vibing on a rainy Sunday afternoon" Acoustic / Indie / Jazz "acoustic chill", "jazz piano", "indie folk" alx search "acoustic chill" --source kw

Search Automation Workflow

Workflow 1: Mood-Based Playlist Generation

  1. Receive user mood request (e.g. "feeling nostalgic").
  2. Generate 3 distinct related query terms (e.g., "90s classic hits", "retro ballad", "childhood memories").
  3. Perform searches for each, fetch the top 2 IDs from each search, and compile a playlist:
# Create temporary mood playlist
alx playlist create "Nostalgic Vibe"

# Add tracks from distinct searches
alx search "周杰伦 经典" --id-only --limit 2 | xargs -I {} alx playlist add "Nostalgic Vibe" {}
alx search "陈奕迅 怀旧" --id-only --limit 2 | xargs -I {} alx playlist add "Nostalgic Vibe" {}

# Play the playlist
alx playlist play "Nostalgic Vibe" --shuffle

Workflow 2: Platform-Optimized Discovery

Different music platforms excel in different genres. The agent should leverage these strengths:

  • NetEase Music (wy): Excellent for indie, lofi, electronic, and user-curated playlist matches.
  • Kuwo (kw) / Kugou (kg): Excellent for high-resolution lossless versions and mainstream Mandarin pop.
  • Migu (mg) / QQ (tx): Excellent for high-fidelity master cuts and exclusive license songs.