Agent Skills for the Mubert Music API v3 — royalty-free AI music generation, streaming, and library access for your apps, games, videos, and other projects.
| Skill | Description |
|---|---|
| mubert-setup | Onboarding wizard — get API key, validate, save to .env, choose single-user or multi-user setup |
| mubert-generate | Generate tracks from text prompts, images, or playlists. Edit instruments and stems. |
| mubert-streaming | Stream real-time AI music with intensity control, loop mode, and restart |
| mubert-library | Browse and filter the pre-made music library by genre, theme, BPM, activity |
| mubert-manage | Manage licenses, customers, and webhooks (service-level administration) |
npx skills add MubertAI/skills- Get your API key at mubert.com/api
- Ask your AI assistant: "Set up Mubert API" — the
mubert-setupskill will walk you through configuration - Start creating: "Generate a 30 second chill lo-fi track"
mubert-setup → configure credentials
↓
mubert-generate → create tracks from prompts/playlists/images
mubert-streaming → stream real-time music in your app
mubert-library → discover tracks by genre, mood, BPM
↓
mubert-manage → manage customers, licenses, webhooks at scale
Copy .env.example to .env and fill in your credentials:
cp .env.example .env| Variable | Required by | Description |
|---|---|---|
MUBERT_COMPANY_ID |
manage, setup (also generate/streaming when minting a customer) | Company identifier (service-level) |
MUBERT_LICENSE_TOKEN |
manage, setup | License token (service-level) |
MUBERT_CUSTOMER_ID |
generate, streaming, library | Customer identifier (public-level) — data.id from POST /service/customers |
MUBERT_ACCESS_TOKEN |
generate, streaming, library | Access token (public-level) — data.access.token from the same response |
MUBERT_NAMESPACE |
multi-user setups | Prefix scoping per-user custom_ids to one app and environment |
The signup email contains only the two service-level values. The public-level pair exists only after a customer is created via the service API — service credentials are rejected by public endpoints.
In a multi-user app the public-level pair is minted per user at runtime and belongs in your database, not in .env — a shared MUBERT_CUSTOMER_ID means every user draws on one customer and one usage counter. MUBERT_NAMESPACE keeps staging from colliding with production: custom_id is unique per company and creation is find-or-create, so two deployments that derive the same custom_id silently share one customer and one token.
Every track generated or streamed through the Mubert API is royalty-free and cleared for commercial use — apps, games, videos, podcasts, social media, and more. No copyright claims, no additional licensing fees.
skills/
├── README.md
├── .env.example
├── mubert-setup/
│ └── SKILL.md
├── mubert-generate/
│ ├── SKILL.md
│ └── references/
├── mubert-streaming/
│ ├── SKILL.md
│ └── references/
├── mubert-library/
│ ├── SKILL.md
│ └── references/
├── mubert-manage/
│ ├── SKILL.md
│ └── references/
├── mubert-audio-montage/
│ └── SKILL.md
├── tests/
│ ├── README.md
│ ├── lint_docs.py
│ └── contract/
└── evals/
├── mubert-setup/
│ ├── evals.json
│ └── trigger_eval.json
├── mubert-generate/
│ ├── evals.json
│ └── trigger_eval.json
├── mubert-streaming/
│ ├── evals.json
│ └── trigger_eval.json
├── mubert-library/
│ ├── evals.json
│ └── trigger_eval.json
└── mubert-manage/
├── evals.json
└── trigger_eval.json
