Best OpenClaw Skills for AI Music Composition
OpenClaw ships a built-in music_generate tool with five audio providers, but ClawHub hosts specialized skills that push music generation further. The built-in tool, Evolink Music, Cynaps3, ACE Music, and claw.fm each target different workflows, from quick background tracks to full production suites with library management.
How We Compared These Skills
OpenClaw gives you two paths to AI music generation. The built-in music_generate tool ships with the platform and connects to cloud providers like Google Lyria and MiniMax. ClawHub skills add specialized capabilities on top, connecting your agent to models like Suno and ACE-Step that the built-in tool does not cover.
We tested five options across both categories. Evaluation criteria included provider coverage (which music models does the skill support?), output control (duration, BPM, key, vocals, and style parameters), pricing model, and installation complexity. We also looked at unique features that set each skill apart, such as library management, batch generation, or monetization.
Here is how the five options compare at a glance:
- Built-in music_generate: Google Lyria 3, MiniMax, fal, OpenRouter, and ComfyUI. No installation needed. Pay provider API fees only.
- Evolink Music: Suno v4, v4.5, and v5 via EvoLink. One-command install. Paid API with model-tier pricing.
- Cynaps3 Plugin: Suno and Sonauto with 26 agent tools. Complex setup (Supabase + Clerk). Lifetime deal pricing.
- ACE Music: ACE-Step 1.5 model. One-command install. Free API with no documented usage limits.
- claw.fm: Autonomous radio station with USDC monetization. Single skill file. Free to submit, revenue from listener tips.
Each entry below follows the same structure: what the skill does, its strengths and limitations, who it is best for, and what it costs.
1. OpenClaw's Built-in music_generate
The music_generate tool shipped with OpenClaw 2026.4.5 as a shared tool available to any agent. It connects to five providers out of the box: Google Lyria 3, MiniMax, fal, OpenRouter, and ComfyUI. Your agent calls it with a text prompt describing the music it wants, and the tool generates an audio file that attaches directly to the agent's reply.
The tool accepts several optional parameters beyond the required prompt. You can pass explicit lyrics, set an instrumental flag for vocals-free output, specify a target duration in seconds, choose between MP3 and WAV output, and override the default provider with a specific model string. Not every provider supports every parameter. Google Lyria 3 ignores duration control and generates whatever length it decides. MiniMax supports duration steering, batch generation, and MP3 output. The tool handles incompatibilities with warnings rather than errors, so your agent does not crash when it hits a gap.
Provider failover is automatic. If the primary provider is unavailable, the tool tries configured fallbacks before returning an error. For longer generation jobs, the tool runs in the background and notifies your agent when the track is ready, keeping conversations responsive.
Key strengths:
- Zero installation, ships with OpenClaw 2026.4.5 and later
- Five providers with automatic failover
- Background task handling keeps conversations responsive
Key limitations:
- No Suno access (requires a third-party skill)
- Parameter support varies across providers
Best for: Agents that need music generation without extra dependencies or API keys beyond the provider.
Pricing: Provider API costs only. Each provider requires its own API key or credentials (Google needs a Gemini key, MiniMax needs its own key, etc.).
How to Choose a ClawHub Music Skill
ClawHub hosts music generation skills that extend OpenClaw beyond the built-in tool. These connect your agent to specialized music models like Suno and offer features the built-in tool does not cover, including vocal control, batch generation, and full library management.
The choice depends on three factors: what model you want access to, how much control you need over the output, and whether you need library management for ongoing production. If your agent generates a few tracks per week for content backgrounds, a simple skill like Evolink Music covers it. If your agent manages a growing catalog with mood tagging, batch exports, and project organization, Cynaps3 justifies its heavier setup.
Consider the credential overhead too. Each skill requires its own API key or service account. Evolink needs one key from evolink.ai. Cynaps3 needs Supabase credentials plus a Clerk user ID. Before committing, check whether your deployment environment can store those secrets securely and rotate them when needed.
2. Evolink Music
Evolink Music connects your OpenClaw agent to Suno's music generation models through a single EvoLink API key. The skill supports five Suno model variants, each balancing quality against generation time and duration limits.
The models range from suno-v4 (the default, good quality, up to 120 seconds) through three v4.5 variants with better quality and 240-second ceilings, up to suno-v5 for studio-grade output at 240 seconds. Two generation modes give you control over the process. Simple mode lets the AI handle everything from a text prompt. Custom mode lets you specify lyrics, vocal gender (male or female), style tags, negative tags to exclude unwanted styles, and a target duration between 30 and 240 seconds.
Installation takes a single command from ClawHub. You need an EvoLink API key from evolink.ai, which you can set by telling your agent directly. The skill supports MP3, WAV, FLAC, AAC, OGG, and M4A output formats with a 100MB file size limit.
One practical consideration: generated file URLs expire. Result URLs last 24 hours and uploaded files last 72 hours. If your agent generates a track and someone requests it two days later, the original link is gone. You will need a persistent storage solution for anything worth keeping.
Key strengths:
- Five Suno models from economical (v4) to studio-grade (v5)
- Fine-grained vocal control with gender, style tags, and negative tags
- One-command install with straightforward API key setup
Key limitations:
- Paid API (EvoLink key required, pricing varies by model tier)
- Result URLs expire in 24 hours
Best for: Production-quality music where vocal control and model selection matter.
3. Cynaps3 OpenClaw Plugin
Cynaps3 goes beyond generation. It is a 26-tool suite covering the full music production lifecycle: creation, library management, project organization, and batch operations. Where other skills stop at "generate a track," Cynaps3 lets your agent build and manage a growing music library over time.
Generation uses dual providers, Suno and Sonauto, with batch generation supporting up to 20 tracks at once. The skill ships with over 150 artist styles across 15 categories including hip-hop, electronic, jazz, K-pop, and classical. Beyond generation, 10 library management tools let agents search existing tracks by mood, genre, energy level, BPM, key, or text. Six project operation tools handle organized collection management, and four write operations cover rating, album creation, energy curve setup, and batch renaming.
Security follows a thin-client architecture. All enforcement happens server-side in Supabase edge functions. Destructive operations require signed confirmation tokens that expire after five minutes. Deleting a project demands triple confirmation with item count verification. No secrets are exposed in the agent runtime.
Setup is heavier than a simple skill install. You need Supabase credentials (URL, anon key, service role key) and a Clerk user ID. Two bundled skill playbooks, cynaps3-core and musicmation, guide the agent through auth protocols and generation decisions.
Key strengths:
- Full production suite with 26 specialized tools
- Library management with search by mood, genre, BPM, key, and energy
- Dual-provider generation with 150+ artist styles and batch support
Key limitations:
- Complex setup requiring Supabase and Clerk configuration
- Higher barrier to entry than single-purpose generation skills
Best for: Agents running ongoing music production workflows that need library management, batch operations, and organized collections.
Pricing: Lifetime deal pricing for early adopters covering current and future modules (Comicmation, Storymation, Skillmation, Contentmation).
Give your music agent persistent file storage
50GB free workspace with MCP access. Store generated tracks, share with collaborators, and hand off finished music to your team. No credit card required.
What Free Options Exist for Agent Music Generation
Not every music generation workflow needs a paid API. ACE Music offers free generation through an open model, while claw.fm flips the economics by letting your agent earn from its compositions.
Free options carry different tradeoffs. ACE Music has no generation costs, but security flags from VirusTotal and OpenClaw's moderation system mean you should audit the skill's network behavior before deploying it alongside sensitive data. Run it in an isolated environment first and monitor outbound requests. claw.fm has no submission costs either, but revenue depends on listener adoption of a crypto-tipping model, which limits near-term earning potential.
For prototyping and experimentation, free skills let you test whether AI music generation fits your agent's workflow before committing to Suno API costs. Once you validate the use case, upgrading to Evolink Music or Cynaps3 is straightforward since OpenClaw skills share the same installation pattern.
4. ACE Music
ACE Music runs on the ACE-Step 1.5 model through a free hosted API. No generation costs and no documented usage limits. The skill offers parameter control that rivals paid alternatives: BPM, musical key, language, duration, and batch output settings. It supports multi-language vocals including Japanese and Korean, plus features beyond basic generation like song covers and section repainting (regenerating a specific portion of a track while keeping the rest intact).
Installation follows the standard ClawHub pattern with a single command. You need a free account at acemusic.ai for an API key, but the API itself charges nothing for generation.
One significant caveat: both VirusTotal and OpenClaw's own moderation system flag this skill as suspicious. The skill functions, but those flags mean you should review its behavior carefully before running it in a production environment or with sensitive data.
Key strengths:
- Completely free API with no documented usage limits
- Detailed parameter control (BPM, key, language, duration)
- Multi-language vocal support and section repainting
Key limitations:
- Flagged as suspicious by VirusTotal and OpenClaw moderation
- Newer model with less community validation than Suno
Best for: Experimentation and prototyping without API costs, after reviewing the security flags.
Pricing: Free.
5. claw.fm
claw.fm takes a different approach. Instead of just generating music, it gives your agent a way to distribute and monetize compositions. The platform runs a 24/7 autonomous radio station where every track is produced by an AI agent. Setup takes one skill file added to your OpenClaw agent.
The economics work through USDC, a stablecoin on the Base blockchain. Listeners can tip tracks or purchase them outright. Revenue splits at 75% to the artist agent, 20% to a shared royalty pool distributed by play count, and 5% to the platform. The codebase is open source on GitHub.
This is early-stage software. The creators have explicitly deferred copyright registration, licensing contracts, and exclusivity definitions to focus on validating core functionality first. If your use case requires legal clarity around music ownership, claw.fm does not provide that yet. The monetization model also depends on listener adoption and comfort with cryptocurrency transactions.
Key strengths:
- Built-in monetization path for agent-generated music
- Autonomous operation after initial setup
- Open source codebase
Key limitations:
- No copyright or licensing infrastructure
- Revenue depends on listener adoption and crypto familiarity
Best for: Experimental agents exploring autonomous music creation and distribution.
Pricing: Free to submit. Revenue from listener tips and purchases via USDC.
How to Pick the Right Skill for Your Workflow
Start with the built-in music_generate if you want to test what AI music generation looks like inside your agent workflow. It requires no installation, supports five providers, and handles failover automatically. Most agents generating occasional background music or audio content can start here with Google Lyria 3 or MiniMax.
When you need higher-quality output or access to Suno models, Evolink Music is the most straightforward upgrade. One install command, one API key, and you get five Suno model tiers with vocal and style control. For agents running sustained music production workflows with growing libraries, Cynaps3 adds library management, batch operations, and project organization that justify its more complex setup.
ACE Music is worth exploring if you want to avoid API costs entirely. The free model and detailed parameter control make it useful for prototyping, though the security flags warrant careful review before deploying in production. And claw.fm is an interesting experiment for anyone building agents that should operate autonomously as music producers with a revenue path.
Storing Generated Music Files
One practical consideration that applies across all these skills: generated music files need a persistent home. The built-in music_generate attaches files to agent replies within the session, but third-party skills often use temporary hosting. Evolink Music result URLs expire in 24 hours. If your agent generates a track on Monday and a collaborator asks for it on Wednesday, the link is dead.
Local disk storage works for single-agent setups on one machine. S3 or Google Drive can store files for teams, but neither indexes audio content or provides agent-friendly access without custom integration work.
Fast.io provides workspaces where agents store, organize, and share generated music through its MCP server. Files uploaded to a workspace are indexed by Intelligence Mode, making them searchable by content and metadata. When a composition is ready for review, the agent can create a branded share link and hand the workspace off to a human collaborator. The free agent plan includes 50GB of storage, 5,000 credits per month, and five workspaces with no credit card required. You can also connect Fast.io to OpenClaw through the storage-for-openclaw integration path.
Frequently Asked Questions
Does OpenClaw have built-in music generation?
Yes. The music_generate tool shipped with OpenClaw 2026.4.5. It supports five providers: Google Lyria 3, MiniMax, fal, OpenRouter, and ComfyUI. Any agent can call it with a text prompt to generate audio, and the result attaches directly to the agent's reply.
What music models work with OpenClaw?
Through the built-in tool and ClawHub skills combined, OpenClaw supports Google Lyria 3, MiniMax, Suno v4/v4.5/v5 (via Evolink Music), Sonauto (via Cynaps3), and ACE-Step 1.5 (via ACE Music). ComfyUI workflows can also route to custom models hosted locally or on Comfy Cloud.
How do I generate a song with OpenClaw?
Describe the music you want in natural language. If the built-in music_generate tool has a configured provider, your agent generates audio directly from your prompt. For Suno-based generation, install the Evolink Music skill from ClawHub, set your EvoLink API key, and prompt your agent to create a track with your desired style, tempo, and vocals.
Can OpenClaw compose instrumental music?
Yes. Both the built-in music_generate tool and ClawHub skills like Evolink Music support an instrumental flag. Set it to true, and the generator produces a vocals-free track. Evolink Music also supports negative tags to exclude specific vocal styles from the output.
Is there a free OpenClaw music generation skill?
ACE Music offers a free hosted API powered by the ACE-Step 1.5 model. You need a free account at acemusic.ai for an API key, but there are no documented generation costs. Note that both VirusTotal and OpenClaw flag this skill as suspicious, so review it before production use.
How long can OpenClaw-generated music tracks be?
Duration depends on the provider and model. Google Lyria 3 does not support explicit duration control. MiniMax supports duration steering. Suno models through Evolink Music range from 120 seconds (v4) to 240 seconds (v4.5 and v5). ACE Music also supports configurable duration through its API parameters.
Related Resources
Give your music agent persistent file storage
50GB free workspace with MCP access. Store generated tracks, share with collaborators, and hand off finished music to your team. No credit card required.