Run your label from Claude.
Labelflow ships a Model Context Protocol server that opens your catalogue, deliverables, communications and analytics to the AI client you already work in. Create releases, check what is blocking, run a catalogue import, without leaving the conversation.
The release-ops layer should not only live behind a UI.
Most label operators already keep an AI client open next to the inbox. Claude, ChatGPT, Cursor. They write replies in it, summarise threads, sketch the week. Then they switch tabs to do the actual work, which usually means clicking through five tools to find the thing they were just thinking about.
Model Context Protocol changes the shape of that. A client that speaks MCP can call tools directly, read live data from your systems and take real actions on your behalf. Labelflow's MCP server makes the AI you already use a working surface on top of your label, not a side panel to paste notes into.
Eighteen tools, two surfaces.
Thirteen read tools and five write tools today, with more landing through the year. Read tools pull live state out of your workspace; write tools act, under the same permission rules as the operator app.
List releases, open one with its tracks and metadata, check status. Create and update releases, update tracks, kick off a catalogue import.
List artists, open a full profile and history, list their tracks across the catalogue. The roster surface of the operator app, readable in conversation.
List comms by artist or release, see what went out, what is drafting and what is waiting on approval. Sends always keep the operator approval gate.
List what is outstanding, blocking or missed, and mark a deliverable done. Read analytics for a release, the catalogue or the label.
The AI you already use becomes a label OS.
Four things operators actually ask on a Monday morning. Pick one.
- Metadata complete · genre missing on track 2
- Distribution sign-off · waiting on you
- Onboarding request · sent 04.08 · replied
- Premaster reminder · sent 12.08 · replied
- Mastering handoff · sent 21.08 · no reply yet
- Artwork approval · drafted 28.08 · waiting on your approval
- DT091 · approved two days late
- DT094 · approved on release day
- DT097 · never marked, live since June
- Top release: Half Light · 22,410
- Strongest new entry: Salt Bloom · 9,830
- Editorial adds: 4 playlists
OAuth, the standard way in.
Click connect inside your AI client, sign in to Labelflow, approve the scopes, done. Every connected app is listed in your settings and can be disconnected in one click.
Everything is logged. Each read and write leaves a trail in the workspace activity log, attributed to the operator and the client, with the arguments it was called with. You can see what the AI did and exactly when.
What MCP will not do.
The same AI policy as the rest of Labelflow. The server inherits every restriction, not only the convenient ones.
No "this artist is about to break" forecasts, here or anywhere else. AI helps you read what exists; the A&R calls stay with you.
MCP cannot send anything today. Its five write tools touch releases, tracks, deliverables and the catalogue import; a future write on communications would carry the same operator approval gate as the rest of Labelflow.
Your workspace data, your client, your decision. Labelflow does not sell aggregated traffic to model labs.
Deletes, status transitions and irreversible moves need an explicit confirmation. An ambiguous prompt cannot quietly empty a release pipeline.
For operators who already think in AI clients.
If you keep Claude open all day, work in Cursor, or run half your label through prompts and copy-paste rituals, this is the version of Labelflow that fits how you already work. If you prefer the operator app, MCP changes nothing for you. It is a parallel surface, not a replacement.
Included on both plans. No add-on cost, on Label and Studio alike.