Connect to RTLCA

Pick your client below. Every client connects to the same hosted endpoint (https://mcp.realtimelca.com/mcp) and authenticates with your Real-Time LCA account via OAuth — no API key, no token to paste.

Using Autodesk Forma or Revit tools? See Connect to RTLCA × Autodesk — it uses a different endpoint and requires an Autodesk account + active Forma hub.

Connecting an external AI client sends your project data to that client and its underlying model provider, so how it’s handled depends on the client you choose. See Transparency & privacy for the full picture across every AI feature.

Connect the Claude web app at claude.ai. Works in any browser — no Claude Desktop install required.

Prerequisites

Claude account
Real-Time LCA account
On Team or Enterprise plans, you need to be an admin to add servers

Open the Connectors page

Go to claude.ai/customize/connectors.

Add a custom connector

Click the + sign → Add custom connector. Paste the URL:

https://mcp.realtimelca.com/mcp

Authenticate the MCP server

Complete any authentication steps Claude prompts for, signing in with your Real-Time LCA account.

Mention RTLCA in a chat

The next time you mention RTLCA in a chat, Claude will detect the connector and ask you for permission to execute tools.

Connect the Claude Desktop app. The flow is the same as Claude.ai — done through the in-app Connectors UI, not a config file.

Prerequisites

Claude Desktop installed (macOS or Windows)
Real-Time LCA account
On Team or Enterprise plans, you need to be an admin to add servers

Open the Connectors page

In Claude Desktop, open Settings → Connectors.

Add a custom connector

Click the + sign → Add custom connector. Paste the URL:

https://mcp.realtimelca.com/mcp

Authenticate the MCP server

Complete any authentication steps Claude prompts for, signing in with your Real-Time LCA account.

Mention RTLCA in a chat

The next time you mention RTLCA in a chat, Claude will detect the connector and ask you for permission to execute tools.

Prerequisites

Node.js ≥ 18
Claude Code installed (npm install -g @anthropic-ai/claude-code)
A Real-Time LCA account

Run the add command

A single command in your terminal:

claude mcp add realtimelca --transport http https://mcp.realtimelca.com/mcp

Authenticate on first use

Claude Code will prompt you to authenticate the first time you use a realtimelca tool. Follow the link to sign in with your RTLCA account; the token is cached for later calls.

Try a first prompt

Start a session and ask, for example:

List all my building projects

Show me the GWP dashboard for the project named "Aalborg Office"

Prerequisites

Codex app
A Real-Time LCA account — you’ll sign in via the browser

Open Codex

Launch the Codex app.

Open the config file

Go to Settings → Configurations → Open config.toml.

Add the rtlca MCP server

Paste the following at the very top of config.toml:

mcp_oauth_callback_port = 5555

[mcp_servers.rtlca]
url = "https://mcp.realtimelca.com/mcp"

Restart Codex

Quit Codex completely — make sure the process is no longer running — then start it again.

Authenticate

Go to Settings → MCP Servers. The rtlca connection should now be listed.Click Authenticate. A browser window opens; sign in with your Real-Time LCA credentials, then return to Codex.

Try a first prompt

Start a chat and ask, for example:

What projects do I have in RTLCA?

Prerequisites

Cursor installed (cursor.com)
Real-Time LCA account

Open Cursor's MCP settings

In Cursor, go to Settings → Tools & MCP and select New MCP Server. mcp.json opens for editing.

Add the realtimelca MCP server

{
  "mcpServers": {
    "realtimelca": {
      "url": "https://mcp.realtimelca.com/mcp"
    }
  }
}

Save the file.

Click Connect to authenticate

Cursor lists the server as Needs authentication. Click the Connect button — your browser opens to the Real-Time LCA sign-in page. Sign in; the token returns to Cursor automatically.

Try a first prompt

Open a new chat and ask, for example:

List all my building projects

Real-Time LCA is published on the ChatGPT App Store — install it in a couple of clicks, no custom connector required.

Prerequisites

A ChatGPT account
A Real-Time LCA account — you’ll sign in via the browser

Open the app listing

Go to the Real-Time LCA MCP Server listing on the ChatGPT App Store, then click Connect in the top right.

ChatGPT App Store listing for the Real-Time LCA MCP Server with a Connect button in the top right. — Real-Time LCA MCP Server on the ChatGPT App Store

Add to ChatGPT

In the Add Real-Time LCA MCP Server to ChatGPT dialog, click Sign in with Real-Time LCA MCP Server.

ChatGPT modal asking the user to add the Real-Time LCA MCP Server, with a Sign in button. — Add Real-Time LCA MCP Server to ChatGPT dialog

A browser window opens to the Real-Time LCA sign-in page. Sign in with your RTLCA credentials, then return to ChatGPT.

Real-Time LCA sign-in page with Email and Password fields. — Real-Time LCA sign-in page

Confirm the connection

ChatGPT confirms with a Real-Time LCA MCP Server is now connected. banner, and the Connect button changes to Start chat.

ChatGPT App Store listing showing a green Real-Time LCA MCP Server is now connected banner and a Start chat button. — Connection confirmation in ChatGPT

Try a first prompt

Click Start chat and ask, for example:

List all my building projects

Manage or review the connection

Open the gear icon next to Start chat on the listing, or go to Settings → Apps → Real-Time LCA MCP Server from anywhere in ChatGPT.

Real-Time LCA MCP Server listing in ChatGPT with a Start chat button and a gear icon to open settings. — Open the connection settings via the gear icon

From here you can toggle Reference memories and chats, review the connection details, or Disconnect the app.

ChatGPT Apps settings showing Real-Time LCA MCP Server with Disconnect, Preferences, and Information sections. — Apps → Real-Time LCA MCP Server settings

Scroll down to Actions for the full list of MCP tools the app exposes and the scopes each one needs.

Actions panel listing rtlca_create_construction, rtlca_create_epd, and other Real-Time LCA MCP tools with their required scopes. — Actions exposed by the Real-Time LCA MCP Server

Create the connector in AI Studio, then attach it to any chat in Le Chat.

Prerequisites

Mistral plan with MCP connector support
Real-Time LCA account

Open AI Studio Connectors

In AI Studio, go to Connectors → Add Connector.

Fill in the connector

Name: any (e.g. Real-Time LCA)
Server URL: https://mcp.realtimelca.com/mcp
Workspace: Private

Click Create.

Creating the connector opens a browser window to the Real-Time LCA sign-in page. Sign in; once successful you’re redirected back to AI Studio.

Use the connector in Le Chat

Start a new chat in Le Chat. Click the + symbol, open the Connectors tab, and select the RTLCA connector. Then prompt, for example:

List all my building projects

Run a local LLM via LM Studio and connect it to the RTLCA MCP server — no Claude or other subscription required, no token limits, no chat data leaving your machine.

Prerequisites

LM Studio installed (lmstudio.ai)
A locally downloaded model that supports tool / function calling
A Real-Time LCA account

Install LM Studio

Download and install LM Studio from lmstudio.ai. Launch the app once installation completes.

Load a local model

In the Discover tab, download a tool-capable model, then load it from the Chat tab’s model selector. Recommended: Gemma 4 or Qwen 3.5 — go larger if your machine can run it. When loading, increase the context size to at least 15k — the MCP tool responses need the headroom.

Add the RTLCA MCP server

Open LM Studio’s Program / MCP panel and edit mcp.json:

{
  "mcpServers": {
    "rtlca-mcp-server": {
      "url": "https://mcp.realtimelca.com/mcp"
    }
  }
}

Save the file. LM Studio will pick up the new server and list its tools.

Click Authenticate in LM Studio

Once the server is in mcp.json, LM Studio shows an Authenticate button on the rtlca-mcp-server entry. Click it — your default browser opens to the Real-Time LCA sign-in page. Sign in with your RTLCA credentials; LM Studio receives the token automatically.

Chat with the connected model

Back in LM Studio’s Chat tab, ask:

List all my building projects

Tool-calling quality varies a lot between local models. If the model ignores the available tools or hallucinates arguments, try a larger variant or one explicitly tuned for function calling.

API documentation

Agent ready

Prerequisites

Prerequisites

Prerequisites

Prerequisites

Prerequisites

Prerequisites

Manage or review the connection

Prerequisites

Prerequisites

​Prerequisites

​Prerequisites

​Prerequisites

​Prerequisites

​Prerequisites

​Prerequisites

​Manage or review the connection

​Prerequisites

​Prerequisites

Prerequisites

Prerequisites

Prerequisites

Prerequisites

Prerequisites

Prerequisites

Manage or review the connection

Prerequisites

Prerequisites