Skip to content

MCP Integration

Your tools. Connected to the task.

An agent is more useful when it can work with the right systems. MCP connects approved tools and data sources without turning every integration into a custom workflow.

Requires a configured MCP server and any server-specific authentication. Local programs and remote endpoints remain separate trust boundaries.

Configure the connection. Keep the controls.

  1. 01

    Choose the server.

    Use a reviewed local program or an approved remote endpoint. Match its access to the project task.

    Add an MCP server.
  2. 02

    Choose its scope.

    Configuration can be global or project-specific. Supported transports include stdio, Streamable HTTP and SSE.

  3. 03

    Review the tool access.

    Server startup approval and permission to call its tools are separate decisions. Do not grant blanket trust just to simplify setup.

Existing Lurus Code screenshot illustrating the product interface and project workflow.
A look at Lurus Code

Runtime visibility

Know what is connected right now.

/mcp shows runtime server status. /mcp refresh-tools explicitly updates the tool registry and visible manifest. A changed server definition is not automatically the same as a refreshed agent session.

Add an MCP server.

List existing entries and add your server from the project directory. The example name local-docs must be unused. Then check connection status in the Lurus Code chat.

This example requires an existing MCP server. Replace the absolute path with its script file. --project saves the configuration in the project; without the flag it is global. add saves the entry. Then start a new CLI session and approve server startup.

Configure local and remote servers
  1. List configured servers in the terminal
    lurus mcp list
  2. Add a local MCP server to the project
    lurus mcp add local-docs --command node --args /absolute/path/server.mjs --project
  3. Inspect runtime status in chat
    /mcp
  4. Inspect runtime status in chat
    /mcp refresh-tools

Your code stays yours.

Source code and request context are processed with Zero Data Retention. No use for model training. Providers and processing regions are disclosed transparently.

See privacy and processing details

A few things to know before you start.

Are only local MCP servers supported?
No. Configuration supports local stdio servers and remote Streamable HTTP or SSE connections. Setup and authentication depend on the server.
Does adding a server give it unrestricted access?
Do not assume so. Server startup and tool execution have separate controls. Review the configured server and the effective tool permissions.
Do privacy guarantees cover every connected service?
No. Model-processing guarantees do not replace a review of the external MCP server and its own data handling.

MCP Integration

Connect one useful tool, deliberately.

Start with a reviewed server and a concrete project task. Check what it can access before letting the agent use it.

Requires a configured MCP server and any server-specific authentication. Local programs and remote endpoints remain separate trust boundaries.

Want to start with your team? Talk to us.

A deliberate choice

Before your first prompt.

Interactive demo
Lurus Codeprivacy-demo · illustrative workspace

Demo session ready. Your model choice and data route are separate decisions.

Illustrative scenario. No real sign-in, security test or code transmission takes place.

Image preview