@cassette.diy/mcp
MCP server for managing Cassette music releases and Creator clips through AI assistants via personal access tokens.
README
@cassette.diy/mcp
MCP server for Cassette (music releases) and Cassette for Creators (short-form clips). Lets AI assistants in Cursor, Claude Desktop, and other MCP clients manage your Cassette account via personal access tokens.
npm: https://www.npmjs.com/package/@cassette.diy/mcp
Docs: https://cassette.diy/docs/mcp
Install
Published on npm. No separate install step when using Cursor — npx downloads the package on first run:
npx -y @cassette.diy/mcp
- A Cassette or Creator account with Google Drive linked (required for uploads).
- A personal access token from the web app:
- Music app: Settings → API / AI access
- Creator app: Settings → API / AI access
Cursor setup
Add to your Cursor MCP config (~/.cursor/mcp.json or project .cursor/mcp.json):
{
"mcpServers": {
"cassette": {
"command": "npx",
"args": ["-y", "@cassette.diy/mcp"],
"env": {
"CASSETTE_API_URL": "https://api.cassette.diy",
"CASSETTE_API_TOKEN": "ct_live_YOUR_TOKEN_HERE",
"CASSETTE_PRODUCT": "both"
}
}
}
}
You can copy a ready-made snippet from Settings after creating a token.
Environment variables
| Variable | Required | Description |
|---|---|---|
CASSETTE_API_URL |
No | Backend URL (default: http://localhost:8080) |
CASSETTE_API_TOKEN |
Yes | Personal access token (ct_live_…) |
CASSETTE_PRODUCT |
No | music, creator, or both (default: both) — controls which tools are registered |
Local development
From the monorepo:
cd cassette-mcp
npm install
npm run build
Point Cursor at the local package:
{
"mcpServers": {
"cassette": {
"command": "node",
"args": ["/absolute/path/to/Cassette/cassette-mcp/dist/index.js"],
"env": {
"CASSETTE_API_URL": "http://localhost:8080",
"CASSETTE_API_TOKEN": "ct_live_...",
"CASSETTE_PRODUCT": "both"
}
}
}
}
Tool catalog
Set CASSETTE_PRODUCT to music, creator, or both (default: both).
Shared
| Tool | Description |
|---|---|
cassette_get_account |
Profile, Drive link status, tiers |
cassette_start_gdrive_link |
OAuth URL to link Google Drive |
cassette_search_users |
Find users for invites (creator mode only) |
Music — track creation workflow
Focused tools for musicians: create a release end-to-end (no delete, collaboration, or Drive sync).
| Tool | Description |
|---|---|
cassette_create_track |
Create track (name, artist, release date) |
cassette_upload_track_file |
Upload artwork, audio, promo, footages, vertical videos |
cassette_update_track_metadata |
Metadata, Spotify pitch, notes |
cassette_list_track_files |
List uploaded files by category |
cassette_get_track |
Get full track after creation |
Creators
Full clip workflow (unchanged):
| Tool | Description |
|---|---|
cassette_get_creator_billing |
Tier, limits, clip count |
cassette_list_clips |
List clips |
cassette_get_clip / create_clip / update_clip / delete_clip |
Clip CRUD |
cassette_update_scenario |
Scenario time blocks |
cassette_update_clip_notes |
Freeform notes |
cassette_list_trends / create_trend / delete_trend |
Trend library |
cassette_start_clip_from_trend |
New clip from trend |
cassette_list_footages / upload_footage / review_footage / delete_footage |
Footages |
cassette_get_final / upload_final / update_post_links |
Final video & links |
cassette_manage_clip_shares |
Collaboration |
Uploads
Upload tools read files from local_path on the machine running the MCP server (your computer when using stdio in Cursor).
- < 10 MB — direct multipart upload
- 10–20 MB — async upload + task polling
- > 20 MB (tracks) — chunked upload + task polling
If Drive is not linked, upload tools fail; use cassette_start_gdrive_link first.
Token scopes
| Scope | Access |
|---|---|
music:read |
GET on /tracks/* |
music:write |
Mutations on /tracks/* |
creator:read |
GET on /creator/* |
creator:write |
Mutations on /creator/* |
Tokens cannot access admin, Stripe checkout, analytics, or token management endpoints.
Publish checklist
- [ ] Deploy backend with
api_tokenscollection and routes - [ ] Verify token create/list/revoke in staging Settings UI
- [ ] Smoke-test MCP tools against staging API
- [x] Publish
@cassette.diy/mcpto npm (v0.1.0) - [ ] Set production
CASSETTE_API_URLin docs - [ ] Optional: track
mcp_token_createdanalytics event
License
UNLICENSED — internal Cassette package.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。