Playlistgen

What This Skill Does

PlaylistGen is a powerful local music management and curation agent designed for OpenClaw. It transforms your static, local directory of audio files (MP3, FLAC, M4A, etc.) into an interactive, AI-driven listening experience. By scanning your files, the skill builds a SQLite database and leverages advanced LLMs—such as Anthropic's Claude Haiku—to tag your music with semantic metadata like mood, energy levels, and sub-genres. Instead of manually organizing files, you interact with your library using natural language, allowing the AI to construct curated, context-aware playlists.

Installation

To get started, first clone the repository into your OpenClaw skills directory: git clone https://github.com/asriverwang/playlistgen ~/.openclaw/skills/playlistgen. Next, navigate to the folder and establish your isolated Python environment using python3 -m venv venv and pip install requests openai anthropic mutagen. Ensure ffprobe (part of the ffmpeg suite) is installed on your system to enable audio metadata extraction. Finally, populate your .env file with your music directory path and your chosen API keys, and you are ready to index your library for the first time.

Use Cases

This skill is perfect for music enthusiasts with large local collections who struggle to find the right "vibe" for specific moments. It is ideal for creating playlists for social events, deep work sessions, or late-night study routines without needing a paid cloud subscription. Because it runs locally, your listening data remains private, and you maintain complete control over your music files.

Example Prompts

1. "I'm hosting a dinner party. Create a playlist of upbeat but mellow jazz and soul tracks from the 1960s and 70s."
2. "Generate a high-energy, late-night driving playlist with 80s synth-pop and modern darkwave that keeps the tempo high."
3. "Show me all the obscure indie tracks in my library that have a melancholic mood and acoustic instrumentation."

Tips & Limitations

For the best results, ensure your directory is well-organized before indexing. Since the LLM uses metadata to reason about your music, ensure your tags are accurate. Note that the initial indexing process may take time depending on your library size and internet speed, as each song is sent for semantic enrichment via the API. Always review the MUSIC_RULES.md file in the repository to understand how the LLM interprets your requests and how it handles constraints like track diversity or artist repetition.