← 返回 Skills 市场

Music Generation

Name: Music Generation
Author: ivangdavila

作者 Iván · GitHub ↗ · v1.0.0

linuxdarwinwin32 ⚠ suspicious

3134

总下载

当前安装

版本数

在 OpenClaw 中安装

/install music-generation

功能描述

Generate AI music with optimized prompts, style control, and production-ready audio output.

使用说明 (SKILL.md)

AI Music Generation

Help users create AI-generated music and audio.

Rules:

Ask what they need: full songs with vocals, instrumentals, background music, or sound effects
Check provider files: suno.md, udio.md, stable-audio.md, musicgen.md, mubert.md, soundraw.md, riffusion.md, replicate.md
Check prompting.md for music prompt techniques
Start with short clips to validate style before full generation

Provider Selection

Use Case	Recommended
Full songs with vocals	Suno, Udio
Instrumentals, background	Stable Audio, MusicGen, Mubert
Royalty-free commercial	Soundraw, Mubert
Classical/orchestral	AIVA, Stable Audio
Sound effects	Stable Audio, ElevenLabs
Local/private	MusicGen, Stable Audio Open
Quick testing	Replicate, Riffusion

Prompting Fundamentals

Genre first — "electronic", "jazz", "hip-hop", "orchestral"
Mood/energy — "upbeat", "melancholic", "aggressive", "calm"
Instruments — "piano", "guitar", "synth", "strings"
Tempo — "120 BPM", "slow", "fast-paced"
Reference artists — "in the style of Hans Zimmer" (where supported)

Output Formats

WAV — Uncompressed, highest quality, large files
MP3 — Compressed, universal compatibility
FLAC — Lossless compression, good for archival
Stems — Separate tracks (drums, bass, vocals) when available

Common Workflows

Background Music for Video

Determine video length and mood
Generate instrumental at matching duration
Adjust tempo to match cuts if needed
Mix levels appropriately

Full Song Production

Write or generate lyrics
Describe musical style in detail
Generate multiple variations
Select best, extend or edit
Export stems if available for mixing

Sound Design

Describe sound effect clearly
Specify duration needed
Generate variations
Layer and process as needed

Licensing Considerations

Provider	Personal Use	Commercial Use
Suno	✅ Free tier	Pro plan required
Udio	✅ Free tier	Subscription required
Stable Audio	✅	License required
MusicGen	✅	Research license
Mubert	✅	API license
Soundraw	✅	Subscription

Always check current licensing terms before commercial use.

Quality Tips

Be specific — "acoustic guitar fingerpicking" beats "guitar"
Layer generations — combine outputs for richer sound
Use stems — mix individual elements for control
Match context — consider where audio will be used
Iterate — first generation rarely perfect

Current Setup

\x3C!-- Provider: status -->

Projects

\x3C!-- What they're creating -->

Preferences

\x3C!-- Preferred styles, providers, settings -->

Check provider files for detailed setup and API usage.

安全使用建议

This skill is a curator/guide for many music-generation providers and appears coherent for that purpose, but proceed cautiously: - Expect to provide API keys/tokens for providers you want to use (Replicate, Stability, Mubert, Soundraw, etc.). The skill's registry entry does not declare these env vars — ask the publisher which credentials it needs and how they are handled. - Some provider examples reference unofficial third‑party endpoints (e.g., Suno unofficial APIs, PiAPI). Verify the authenticity and privacy policies of any non-official endpoints before sending content or keys; prefer official vendor APIs. - Never paste high‑privilege secrets into chat. If you must provide keys, create scoped, limited API keys with usage/financial limits and revoke them after testing. - The skill may make network calls when invoked. If you want to limit risk, disable autonomous invocation for this skill or monitor outgoing requests/logs until you trust it. - For commercial use, double-check licensing terms for each model/provider (several entries note research or subscription licenses). If you want to install this skill, ask the maintainer to (1) list exactly which env vars/credentials are required, (2) indicate which endpoints are official, (3) explain whether the skill stores any tokens or sends data to third parties, and (4) provide an option to run in a local-only mode (e.g., MusicGen/Stable Audio local models) to avoid sending material to external services.

功能分析

Type: OpenClaw Skill Name: music-generation Version: 1.0.0 The skill bundle consists entirely of documentation files (`.md`) and a metadata file (`_meta.json`). The `SKILL.md` provides instructions for the AI agent that are aligned with the stated purpose of music generation, without any evidence of prompt injection attempts to deviate from this purpose. The other `.md` files contain legitimate Python code snippets demonstrating interaction with various music generation APIs (Mubert, Replicate, Stability AI, Soundraw) and local libraries (MusicGen, Stable Audio Open, Riffusion). File I/O operations are limited to reading/writing audio files, which is central to the skill's function. While some documentation mentions unofficial APIs for Suno and Udio, this is presented as information for the user rather than an instruction to engage in malicious activity. There is no evidence of data exfiltration, unauthorized execution, persistence mechanisms, or obfuscation.

能力评估

⚠ Purpose & Capability

The skill is legitimately a multi-provider music-generation guide and the included provider files (MusicGen, Stable Audio, Suno, Replicate, Mubert, Soundraw, Riffusion, Udio) align with the stated purpose. However, the registry metadata declares no required environment variables or credentials, while the provider files contain numerous examples using API keys and tokens (e.g., API_KEY, REPLICATE_API_TOKEN, STABILITY_API_KEY). That mismatch (no declared secrets but many example credentials) is disproportionate and unexplained. The SKILL.md also references unofficial APIs/wrappers for providers that state they have no official API (e.g., Suno unofficial endpoints), which deserves scrutiny.

ℹ Instruction Scope

The runtime instructions stay focused on music generation and prompting best practices. They direct the agent to consult the provider files for API usage. The provider files include code samples that perform network calls, open local files (e.g., open('melody.wav','rb')), poll webhooks, and show webhook URLs — all reasonable for a multi-provider integration. The instructions do not ask the agent to read arbitrary system secrets or unrelated files, but because they instruct the agent to follow provider usage, the agent could be directed to send user content or keys to external endpoints if invoked.

✓ Install Mechanism

This is instruction-only with no install spec and no code files that run on install — lowest-risk install mechanism. Nothing is written to disk by an installer. The provider docs reference installing provider SDKs (pip installs) for local models, which is normal and expected, but these are not performed automatically by the skill.

⚠ Credentials

Although the skill metadata lists no required env vars, the provider files repeatedly reference multiple credentials (REPLICATE_API_TOKEN, STABILITY_API_KEY, generic API_KEY placeholders, Bearer tokens, webhook URLs). A user would need to supply multiple unrelated API keys to use the integrations. The SKILL.md does not declare or scope these credentials (no primaryEnv or required.env), so it's unclear how secrets are expected to be provided, stored, or used — this is a proportionality and transparency problem. Some provider entries point to unofficial third‑party APIs/wrappers (Suno via api.sunoapi.org, piapi.ai) which increase risk of credential leakage or unexpected data handling if used.

✓ Persistence & Privilege

The skill does not request persistent/always-on presence (always: false). Autonomous invocation is permitted by default (disable-model-invocation: false), which is expected for skills. There is no evidence the skill requests to modify other skills or system-wide configuration. If combined with the environment concerns, consider restricting autonomous use until credentials/endpoints are verified.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install music-generation
安装完成后，直接呼叫该 Skill 的名称或使用 /music-generation 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

Initial release

元数据

Slug music-generation

版本 1.0.0

许可证 —

累计安装 15

当前安装数 15

历史版本数 1

常见问题