Magic Master MCP
# Magic Master MCP — audio mastering for AI agents
Remote [MCP](https://modelcontextprotocol.io) server that lets Claude, ChatGPT, Cursor, VS Code, Claude Code and any other MCP client **master audio**: hit a LUFS target with True Peak limiting, strip the **Suno / Udio AI fingerprint**, and get a mastering passport back in the chat.
- Server: `https://magicmaster.pro/mcp` (streamable HTTP, protocol `2026-07-28`, also `2025-11-25` / `2025-06-18`)
- OAuth endpoint for paid tools: `https://magicmaster.pro/mcp/oauth`
- Registry entry: [`pro.magicmaster/mastering`](https://registry.modelcontextprotocol.io/v0/servers?search=magicmaster)
- Agent playbook: <https://magicmaster.pro/agents.md> · machine manifest: <https://magicmaster.pro/agent.json>
- Human page (RU/EN): <https://magicmaster.pro/mcp>
Nothing to install: the server is hosted. This repository holds the connection recipes and a tiny stdio bridge for clients that cannot speak HTTP.
## One-click install
[](cursor://anysphere.cursor-deeplink/mcp/install?name=magicmaster&config=eyJ1cmwiOiJodHRwczovL21hZ2ljbWFzdGVyLnByby9tY3AifQ==)
[](vscode:mcp/install?%7B%22name%22%3A%22magicmaster%22%2C%22type%22%3A%22http%22%2C%22url%22%3A%22https%3A%2F%2Fmagicmaster.pro%2Fmcp%22%7D)
## Connect
**Claude Code**
```bash
claude mcp add --transport http magicmaster https://magicmaster.pro/mcp
```
**claude.ai / Claude Desktop (custom connector, OAuth):** Settings → Connectors → *Add custom connector* → URL `https://magicmaster.pro/mcp/oauth` → sign in on the consent page. Use `https://magicmaster.pro/mcp` if you only need the free tools.
**Claude Desktop (`claude_desktop_config.json`)**
```json
{ "mcpServers": { "magicmaster": { "type": "http", "url": "https://magicmaster.pro/mcp" } } }
```
**Cursor (`.cursor/mcp.json`)**
```json
{ "mcpServers": { "magicmaster": { "url": "https://magicmaster.pro/mcp" } } }
```
**VS Code (`.vscode/mcp.json`)**
```json
{ "servers": { "magicmaster": { "type": "http", "url": "https://magicmaster.pro/mcp" } } }
```
**ChatGPT (Developer mode → Connectors → Create):** URL `https://magicmaster.pro/mcp/oauth`, authentication OAuth.
**stdio-only clients** — bridge through [`mcp-remote`](https://www.npmjs.com/package/mcp-remote):
```json
{ "mcpServers": { "magicmaster": { "command": "npx", "args": ["-y", "mcp-remote", "https://magicmaster.pro/mcp/oauth"] } } }
```
or the npm package — `npx -y magicmaster-mcp` ([magicmaster-mcp on npm](https://www.npmjs.com/package/magicmaster-mcp)), a one-line wrapper around the same bridge:
```json
{ "mcpServers": { "magicmaster": { "command": "npx", "args": ["-y", "magicmaster-mcp"] } } }
```
## Docker
For clients without Node, and for the Glama build that runs security checks:
```bash
docker build -t magicmaster-mcp .
docker run -i --rm magicmaster-mcp
```
The image holds only the bridge — the server itself stays hosted at
`https://magicmaster.pro/mcp`. It is built from this repository (`npm ci`), so
`mcp-remote` is baked in and the container needs no registry at run time — only
outbound HTTPS to `magicmaster.pro`. The image defaults to the anonymous
endpoint (free tools, no sign-in); switch it to the OAuth endpoint, which also
unlocks the paid tools, with
`-e MAGICMASTER_MCP_URL=https://magicmaster.pro/mcp/oauth`.
Build spec for a Glama release (the platform configures the recipe on its side,
it does not read this Dockerfile):
- build steps: `npm ci --omit=dev`
- start command: `node bin/magicmaster-mcp.js`
- environment: `MAGICMASTER_MCP_URL=https://magicmaster.pro/mcp`
## Tools
| Tool | What it does | Cost |
|---|---|---|
| `analyze_track` | LUFS, true peak, duration, correlation, genre hint | free |
| `clean_ai_trace` | removes the Suno / Udio digital fingerprint, loudness untouched | free |
| `master_track` | mastering: 24 presets, LUFS target, 6 export formats | 1 token |
| `get_job` | status + result metrics, with a mastering passport widget (MCP Apps) | free |
| `check_limits` | balance, remaining free quota, reset time | free |
| `list_presets`, `get_pricing`, `get_service_info` | presets with target loudness, live prices, service manifest | free |
| `create_topup_link` | prepares a checkout for the human — the agent cannot pay by itself | free |
Typical flow: `analyze_track` → (`clean_ai_trace` for AI-generated tracks) → `master_track` → poll `get_job`. A master takes 20–60 s.
## Access and pricing
- Free tools need no account.
- Mastering: 3 per month per IP without an account, 1 per day with a free account.
- Register an agent once (`POST /api/agents/register`) and get **3 trial masters**; afterwards 1 token = 1 full PRO master. Tokens never expire. Live prices: `GET /api/tokens/packages`.
- Rules for agents, error codes and the payment protocol: <https://magicmaster.pro/agents.md>.
## Links
Terms <https://magicmaster.pro/terms> · Privacy <https://magicmaster.pro/privacy> · Public OpenAPI <https://magicmaster.pro/openapi-public.json> · Support support@magicmaster.pro
---
### По-русски
Удалённый MCP-сервер Magic Master: мастеринг под целевую громкость (LUFS, True Peak), снятие цифрового следа Suno/Udio, паспорт результата прямо в чате. Ставить ничего не нужно — адрес `https://magicmaster.pro/mcp`. Подробности и подключение по-русски: <https://magicmaster.pro/mcp>.
TDQS
Scored across 11 tools
Most tools are cleanly separated: analyze_track, clean_ai_trace, preview_styles, master_track, and batch_master all have clearly different purposes. The only mild overlap is among get_service_info, get_pricing, and check_limits, all of which touch cost/account state and require careful reading.
Names are uniformly lowercase snake_case and mostly follow a verb_noun pattern like list_presets, analyze_track, and get_job. The only notable deviation is batch_master, which reads more like a compound noun than the verb_noun pattern used by master_track.
Eleven tools is well within the ideal range and each one earns its place in the mastering workflow. The set covers orientation, presets, pricing, limits, analysis, cleanup, preview, single and batch mastering, job polling, and token purchases without filler.
The full workflow is closed: agents can check limits, analyze, preview, master or batch master, poll for results and downloads, and create a top-up link when tokens run out. There are no dead ends in the core lifecycle.