MCP setup

MCP (Model Context Protocol) is an open protocol that enables AI assistants to securely access and interact with external tools and data sources. By connecting Vectorly as an MCP server to your MCP client, you can access Vectorly routines and capabilities directly from your LLM or IDE.

You automatically get a hosted MCP server with Vectorly. The MCP server is available at https://mcp.vectorly.app/mcp.

Vectorly supports both of the popular authentication methods:

• API key: Use a Vectorly Service Token from the Developer Console via an X-Service-Token: <YOUR_SERVICE_TOKEN> request header.

• OAuth login: Authenticate using OAuth flow directly in your MCP client.

Currently, each account has one default project created at registration; all routines in that project are exposed as tools on your MCP server.

Your data are secure. Authentication is the method by which our MCP server, at https://mcp.vectorly.app/mcp, recognizes you and securely exposes the routines you've created. You may choose to allow others to use your MCP server by providing them a token generated from your project. Otherwise, only you can access your MCP server.

In MCP, a tool call is how an AI assistant invokes a capability on a server. Your client sends a standardized call with a tool name and JSON arguments, and the server executes it.

In this context, each routine you define in the Developer Console becomes a tool that your LLM can invoke via Vectorly's MCP. All routines are automatically registered, no extra setup required.