<aside>

Alpha Program

MCP is currently in alpha as we evaluate which use cases to support. During early testing, we found that expanding functionality often led to increased hallucinations and errors from the AI.

Nitro AI addresses this with its agent architecture, which offloads context to a shared workspace. This approach helps maintain accuracy and is not commonly used by most AI systems, that can connect to the MCP.

</aside>

<aside>

Heads up before you start

AI connected to this MCP can make mistakes -- including creating projects or templates that appear to save but fail to load in the Rocketlane UI.

If that happens, you'll need to manually delete those resources. Rocketlane Support won’t be able to recover those projects, tasks, or templates.

Don't use this in live workflows yet.

</aside>

What can the MCP do?

How to connect to the MCP

We recommend connecting the MCP with an LLM like Claude Sonnet 4.6 on the Desktop. The MCP should work on most LLM apps like Cursor, Claude Desktop, Glean etc. If your AI supports lazy tool loading, we recommend enabling it.

image.png

Example admin configuration - Claude Desktop

Ask your Claude admin to create a custom web connector

image.png

Use the URL: https://rocket-mcp.rl-platforms.rocketlane.com/mcp

This should list the connector for all users in your Claude account.

Your end users should head to Settings → Connectors → Rocketlane and Click on Connect. Once the authentication is completed, the AI will be connected to Rocketlane’s MCP.

You can now chat with Rocketlane to create timesheets, while also bringing in context from other sources like a CSV file uploaded in the conversation or a connected Google Calendar MCP.

Creating timesheets in Rocketlane with AI