QUESTION
How do I enable MCP access and configure it for use with an AI agent?
ANSWER
Introduced in Chief Architect Premier X18 Build 28.2.1.14, a Model Context Protocol (MCP) server is available, enabling integration with a development focused AI agent (e.g. Anthropic Claude Code, Open AI Codex, Google Antigravity, GitHub Copilot from Microsoft) via a local client installation.
Before use, the Chief MCP Server setting within the program's Agentic AI Preference section must be enabled.
Chief Architect's MCP implementation consists of the following:
-
ca-mcp-x18: core server used for interaction with Chief Architect.
-
ca-image-utilities-x18: optional server that allows for image manipulation in clients that don't support it (e.g. Anthropic Claude Code). Google API key required.
-
README.md: informational file containing text details relating to the MCP implementation.
-
pyproject.toml: provides the necessary setup instructions for the AI agent.
REQUIREMENTS & CONSIDERATIONS
- Local AI client installation with local MCP server support (desktop applications, CLI's, and IDE's). Web-based clients are not compatible.
- Both Project Management and Auto Save must be enabled to access the full range of features associated with the Chief Architect MCP.
- Chief Architect's core server is a local STDIO server discoverable where Chief Architect Premier is installed; this is not an internet endpoint or cloud enabled service.
The information in this article is not intended to be an endorsement for the products or companies listed, and listing these products and companies is for your information only. We will not provide any support for these products.
Enabling and Configuring the Chief Architect MCP Server
- Navigate to Edit> Preferences
on a Windows system or Chief Architect> Preferences
on a Mac.
- On the General panel of the Preferences dialog that opens, under the Agentic AI section, check the Enable Chief MCP Server (Experimental) box.

- Thoroughly read over the Information dialog that displays, then click OK when you're ready to proceed. Click OK again to confirm the change and close Preferences.

- Another Information dialog will display stating that the program must be restarted for the MCP settings to take effect. Click OK, then restart the program.

- The MCP server(s) can now be configured using your preferred AI agent that's associated with a local client installation. Consider using the following prompt with your AI agent to get started:
A local MCP server is installed with Chief Architect Premier X18. The default installation location is C:\Program Files\Chief Architect\Chief Architect Premier X18\ on Windows or Applications/Chief Architect Premier X18/ on Mac. Locate the MCP server and configure it for use.
- Follow your AI agent's instructions to proceed. Once the ca-mcp-x18 server has been properly configured and your AI agent can successfully connect to it, the MCP tools will be loaded, and your AI agent will be able to interact with Chief Architect.
TROUBLESHOOTING
- If communication has failed, verify that the Enable Chief MCP Server preference within Chief Architect is enabled.
- If discovery or configuration was unsuccessful, you can start fresh by following the instructions below:
-
- Tell the AI client to:
Purge all Chief Architect MCP information.
- Restart the AI client.
- Provide the following prompt:
Search and locate the camcp package associated with Chief Architect Premier X18. It consists of a local, stdio-based Python MCP server. Once located, configure it for use.