Connecting an AI assistant through MCP
How to connect Claude or another AI assistant to your Wlaunch data through the read-only MCP server
What the Wlaunch MCP server is
Wlaunch has a Model Context Protocol (MCP) server. Connect Claude, ChatGPT or another MCP client to it, and your AI assistant can answer questions about your business data in Wlaunch: clients, appointments, schedules, finance and reports. The assistant only reads data and never changes anything.
This article is the short version. Every MCP client, the limits and troubleshooting are covered on the developers page.
Who can use it
The MCP server is part of the Extended license, together with API integrations. To connect, you need an Extended license and a personal access token.
How to get an access token
Access tokens are issued by Wlaunch support. Contact support and ask for a personal access token for your Wlaunch account.
Treat the token like a password:
- it gives access to everything your account can see, across all of your companies;
- keep it out of shared configs, out of git and out of screenshots;
- get a separate token for each person and never share yours;
- if the token leaks or is no longer needed, contact support and it will be revoked.
How to connect your assistant
The server address:
https://mcp.wlaunch.net/mcp
The server expects the token in a standard header on every request:
Authorization: Bearer <access_token>
In the Claude Desktop step below, replace <access_token> with the token support sent you.
Claude Code
Put the token in the WLAUNCH_MCP_TOKEN environment variable, loaded from a secret manager or your system keychain, then add the server from the terminal:
claude mcp add --transport http wlaunch https://mcp.wlaunch.net/mcp \
--header "Authorization: Bearer $WLAUNCH_MCP_TOKEN"
Check that it was added with claude mcp list. Do not type the token itself into the command: the shell keeps every command in its history. Claude Code writes the token into its config in plain text, so treat that file as a credential.
Claude Desktop
The token goes in the Request headers section of the Add custom connector dialog. Anthropic offers that section as a beta to a limited set of organizations: if your dialog does not have it, connect through Claude Code instead. On Team and Enterprise plans only an Owner can add a custom connector, and the header is then shared by everyone in the organization, so use Claude Code there too.
The order of the steps matters: authentication first, then the header.
- Open Customize → Connectors and press Add custom connector.
- Enter
Wlaunchin the Name field andhttps://mcp.wlaunch.net/mcpin the URL field, then press Continue. - Under Authentication, choose No sign-in. The dialog may pick Sign in now on its own: do not accept it, because the sign-in will fail.
- Under Request headers, press Add header, choose
authorizationand enter the valueBearer <access_token>. - Press Add.
Instructions for Codex CLI, ChatGPT and other MCP clients are on the developers page.
How to check it works
Ask your assistant: "What companies can this integration see?"
It should call the wlaunch_company_context tool and list your companies. Every other question starts there: the tool returns your company's own vocabulary (appointment statuses, branches, tags, finance operation types), which the other tools need in order to answer correctly.
What the assistant can and cannot do
- It reads through 25 tools: clients, appointments, schedules, services, memberships, orders, finance, payroll, inventory, ratings, surveys, messages, and the P&L, cash-flow, cash-balance, statistics and retention reports.
- It only reads. No tool creates, edits, cancels or deletes anything: the assistant will not book an appointment, change a price or add a client.
- It sees no more than you do. Every request is made as your account, with the same access rules as the back office. If your account cannot see a branch's payroll, neither can your assistant.
Example questions for owners and managers are collected in a separate guide.