tapir-archicad-mcp 0.2.0

An MCP server that enables AI agents to automate Archicad using natural language via intelligent, semantic tool discovery.

Archicad Tapir MCP Server

This project provides a Model Context Protocol (MCP) server for Archicad. It acts as a bridge, allowing AI agents and applications (like Claude for Desktop) to interact with running Archicad instances by wrapping the powerful Tapir JSON API.

The server dynamically generates a comprehensive set of over 80 MCP tools from the Tapir API schema, enabling fine-grained control over Archicad projects.

Disclaimer: This project is in an early stage of development. It has not been extensively tested and is intended primarily for experimental and educational purposes. Interfaces and functionality may change in future updates. Please use with caution.

Key Features

• Intelligent Tool Discovery: The server exposes a simple discover_tools function that uses a powerful local semantic search engine to find the most relevant Archicad command from a user's natural language query.
• Massive Toolset, Minimal Footprint: Provides access to the entire Tapir API (80+ commands, expanding to 160+) without overwhelming the AI's context window.
• 100% Local & Private Search: The semantic search index is built and runs entirely on your machine using sentence-transformers and faiss-cpu. No data ever leaves your computer, and no API keys are required.
• Adaptive & Relevant Results: Search uses a sophisticated "Top-Score Relative Threshold" to filter out noise and return only the most relevant tools for a given query.
• Multi-Instance Control: Connect to and manage multiple running Archicad instances simultaneously.
• Robust & Packaged: Designed as a proper Python package with a pyproject.toml, enabling simple and reliable installation.

Installation & Setup

Follow these steps to get the server running and connected to an MCP client like Claude for Desktop.

1. Prerequisites

• Python 3.12+ and uv: Ensure you have a modern version of Python and the uv package manager installed. You can install uv with pip install uv.
• Archicad & Tapir Add-On: You must have Archicad running with the Tapir Archicad Add-On installed. The server cannot function without it.
• MCP Client: An application that can host MCP servers, such as Claude for Desktop.

2. Clone the Repository

Get the project code on your local machine:

git clone https://github.com/your-username/archicad-mcp-server.git
cd archicad-mcp-server

3. Install Dependencies

Create a virtual environment and install the required Python packages using uv.

# Create and activate the virtual environment
uv venv

# On macOS/Linux
source .venv/bin/activate
# On Windows
.venv\Scripts\activate

# Install dependencies from pyproject.toml
uv sync

4. Configure Claude for Desktop

Finally, tell Claude how to run your server. Open your claude_desktop_config.json file and add the following configuration.

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "Tapir_Archicad": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "/path/to/your/archicad-mcp-server",
        "server.py"
      ]
    }
  }
}

Important: You must replace "/path/to/your/archicad-mcp-server" with the full, absolute path to the directory where you cloned the project.

Usage

1. Restart Claude for Desktop to apply the configuration changes.

2. Ensure at least one instance of Archicad (with Tapir) is running.

3. The client will now have access to a small set of core tools. Start by asking it to find the running Archicad instances:

   "Can you check what Archicad projects I have running?"

   The AI will run discovery_list_active_archicads and report the active instances and their port numbers.

4. Now, state your main goal. For example:

   "Okay, using port 12345, get all the Wall elements from the project."

5. The AI will now perform the two-step discover/call workflow:

   • First, it will call archicad_discover_tools with a query like "get all wall elements". The server's semantic search will find that the best match is the elements_get_elements_by_type tool.
   • Second, it will call archicad_call_tool, using the name="elements_get_elements_by_type" it just discovered and constructing the necessary arguments (including the port and params with elementType="Wall").
   • The final result is returned to you.

How It Works

The server operates through a layered architecture:

• AI Agent (e.g., Claude): Interacts with the user and decides which tools to call.
• MCP Client (e.g., Claude for Desktop): Manages the server process and communication.
• Archicad Tapir MCP Server (This Project): Provides an intelligent abstraction layer over the Tapir API, exposing a simple discover/call interface.
• multiconn_archicad Library: The underlying Python library that handles the low-level communication with Archicad instances.
• Archicad & Tapir Add-On: The final destination that executes the commands.

Contributing

Contributions are welcome! Please feel free to submit an issue or open a pull request.

License

This project is licensed under the MIT License. See the LICENSE file for details.