Chronulus MCP Conduit for Claude

Engage with Chronulus AI's Forecasting & Projection Agents within the Claude Environment

Initial Setup: Claude Desktop Client

Installation

The Claude Desktop application is presently accessible on both macOS and Windows platforms.

Obtain the Claude Desktop installer here

Configuration Steps

Adhere to the standard introductory guidelines here for configuring your Claude desktop application instance.

Your specific Claude configuration file resides in one of these system locations:

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

Subsequently, select one of the subsequent setup methodologies that best aligns with your requirements and incorporate it into your claude_desktop_config.json file:

Leveraging pip

(Method 1) Deploy the official release from the Python Package Index (PyPI)

pip install chronulus-mcp

(Method 2) Install directly from the GitHub repository source

git clone https://github.com/ChronulusAI/chronulus-mcp.git
cd chronulus-mcp
pip install .

{
  "mcpServers": {
    "chronulus-agents": {
      "command": "python",
      "args": ["-m", "chronulus_mcp"],
      "env": {
        "CHRONULUS_API_KEY": "<YOUR_CHRONULUS_API_KEY>"
      }
    }
  }
}

**Note:** Should you encounter an error such as "MCP chronulus-agents: spawn python ENOENT", it strongly suggests that the system cannot locate the executable. In such cases, specify the unambiguous, absolute path to your Python interpreter. For instance, use `/Library/Frameworks/Python.framework/Versions/3.11/bin/python3` instead of the unqualified command `python`.

Employing docker

This approach establishes a Docker image, named 'chronulus-mcp', for consistent, reusable deployment within your Claude configuration.

git clone https://github.com/ChronulusAI/chronulus-mcp.git
cd chronulus-mcp
 docker build . -t 'chronulus-mcp'

Ensure that the final argument specified in your Claude configuration exactly matches the tag assigned to the Docker image during the build process.

{
  "mcpServers": {
    "chronulus-agents": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "-e", "CHRONULUS_API_KEY", "chronulus-mcp"],
      "env": {
        "CHRONULUS_API_KEY": "<YOUR_CHRONULUS_API_KEY>"
      }
    }
  }
}

Utilizing uvx (Universal Version eXecutor)

`uvx` facilitates the automated retrieval of the most current `chronulus-mcp` release from PyPI, its subsequent installation, and immediate execution.

{
  "mcpServers": {
    "chronulus-agents": {
      "command": "uvx",
      "args": ["chronulus-mcp"],
      "env": {
        "CHRONULUS_API_KEY": "<YOUR_CHRONULUS_API_KEY>"
      }
    }
  }
}

**Note:** If an error like "MCP chronulus-agents: spawn uvx ENOENT" occurs, you must either: 1. [Install uv](https://docs.astral.sh/uv/getting-started/installation/) on your system, or 2. Supply the fully qualified path to the `uvx` binary. Example: `/Users/username/.local/bin/uvx` rather than just `uvx`.

Complementary Servers (Filesystem Access, Data Fetching, etc.)

Our demonstration utilizes external servers such as fetch and filesystem.

Refer to the respective server maintainer's documentation for detailed instructions on incorporating and configuring these auxiliary servers.

The following configuration snippet illustrates how to integrate filesystem and fetch capabilities alongside the Chronulus service within your claude_desktop_config.json:

{
  "mcpServers": {
    "chronulus-agents": {
      "command": "uvx",
      "args": ["chronulus-mcp"],
      "env": {
        "CHRONULUS_API_KEY": "<YOUR_CHRONULUS_API_KEY>"
      }
    },
    "filesystem": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-filesystem",
        "/path/to/AIWorkspace"
      ]
    },
    "fetch": {
      "command": "uvx",
      "args": ["mcp-server-fetch"]
    }
  }
}

Claude Interaction Preferences

To optimize your workflow when utilizing Claude with diverse toolsets, it is advisable to embed your specific instructions within the main Claude Settings area.

You can revise your Claude configuration settings through two primary avenues:

Via Claude Desktop: Navigate to Settings -> General -> Claude Settings -> Profile (tab)
Via the web interface: Access claude.ai/settings and select the Profile (tab)

These defined preferences are synchronized across both the Claude Desktop application and the web-based Claude interface, meaning your directives must remain functional across both execution environments.

The prescribed preferences utilized in our demonstration examples are detailed below:

## Tool-Dependent Protocols
The subsequent directives apply exclusively when MCP Servers are actively reachable.

### Filesystem Access Tool Directives
- Refrain from invoking 'read_file' or 'read_multiple_files' operations on proprietary binary file formats (e.g., image assets, PDF documents, Word documents).
- When interacting with non-textual binary assets (e.g., images, PDFs, docx), utilize the 'get_info' function for preliminary inspection instead of the 'read_*' set of operations.

### Chronulus Agents Tool Directives
- When invoking Chronulus functionalities, prioritize supplying inputs via specialized field types such as TextFromFile, PdfFromFile, and ImageFromFile over attempting direct ingestion/scanning of file contents.
- If Chronulus generates a visual forecast representation (a plot), you must always append the narrative explanation provided by Chronulus directly beneath the visualization, explicitly labeling this section as "Chronulus Explanation."