Music Skill

Search songs, download playable audio, fetch lyrics, parse music share links, configure platform cookies, and switch music sources through a local go-music-api backend on Linux, macOS, and Windows. Use when the user asks to search music, play or download a song, get lyrics, resolve music share links, configure cookies for login-required tracks, or retry playback from another source after failures. If the backend is missing or unhealthy, install and start go-music-api first.

go-music-api

Use this skill to install and run a local go-music-api backend, search tracks across sources, download audio, embed metadata, and recover from source failures.

Primary workflow

Prefer the bundled scripts instead of reimplementing the flow by hand.

Route by platform

• On Linux or macOS, use scripts/install.sh and scripts/play.sh.
• On Windows, read docs/windows.md before proceeding.
• When the user mentions cookies, VIP-only tracks, grey tracks, or login-required tracks, read docs/cookies.md.

Linux/macOS install

Run:

scripts/install.sh

The install script should:

• install go-music-api into ~/.openclaw/music
• choose a usable local port
• start the backend in the background
• verify health with a local API request

Linux/macOS download

Run:

scripts/play.sh "稻香" "$HOME/.openclaw/media/daoxiang.mp3"

The play script should:

• search by query
• handle nested search payloads such as data.data, data.list, or data.songs
• rank candidates by song title, artist, and source quality
• avoid karaoke, cover, remix, live, DJ, and instrumental variants when possible
• download the audio stream to the requested path
• reuse cached files when an equivalent file already exists
• call scripts/embed_metadata.py to write title, artist, album, cover art, and embedded lyrics when available

Prefer saving final media under a sendable location such as ~/.openclaw/media/.

Manual API workflow

Use this only for debugging or when the helper scripts need changes.

1. Ensure the backend is installed and running.
2. Read ~/.openclaw/music/port on Linux/macOS or %USERPROFILE%\.openclaw\music\port on Windows; default to 8080 if absent.
3. Search with GET /api/v1/music/search?q={q}.
4. Parse list results from the top-level response or nested data.* collections.
5. Choose the best candidate.
6. Download audio with GET /api/v1/music/stream?id={id}&source={source}.
7. Treat the stream response as audio bytes, not JSON.
8. If playback fails, try GET /api/v1/music/switch?... to switch source and retry.
9. If user provides account cookies, read docs/cookies.md, set them with POST /api/v1/system/cookies, and verify with GET /api/v1/system/cookies.
10. Fetch lyrics with GET /api/v1/music/lyric?id={id}&source={source} when needed.

Files and state

Runtime files live under ~/.openclaw/music (Linux/macOS) or %USERPROFILE%\.openclaw\music (Windows):

• binary: go-music-api (Linux/macOS) or go-music-api.exe (Windows)
• log: log.txt
• pid: pid
• port: port
• cache index: cache-index.json

Failure handling

• If installation fails, check platform and architecture detection, GitHub Releases reachability, and required tools such as curl, tar, unzip, and file (Linux/macOS) or the Windows requirements in docs/windows.md.
• Match release asset names exactly. Do not use loose matching that could select .deb or .rpm packages.
• Accept only native executables after extraction. Fail immediately for text files, HTML, scripts, or package files. On Windows, validate the PE header (MZ signature).
• If metadata embedding is required, ensure Python and mutagen are available. If not, skip metadata embedding or install the dependency before retrying.
• If certain tracks fail due to platform restrictions, ask for platform cookies and apply them via /api/v1/system/cookies before switching sources.
• If startup health checks fail, inspect the runtime log.txt.
• If the backend always binds to a fixed port in practice, simplify the port logic instead of pretending dynamic ports work.