Chronulus MCP

What ChronulusAI/chronulus-mcp MCP server does

ChronulusAI/chronulus-mcp MCP server exposes Chronulus AI forecasting and prediction agents through the Model Context Protocol. Its purpose is to let a supported MCP client, especially Claude Desktop, interact with Chronulus agents during a conversation. The repository description focuses on predicting values and outcomes with Chronulus rather than implementing a forecasting model inside the local process.

The available material does not provide a formal list of MCP tool names or schemas. It does show usage guidance for Chronulus agent workflows: prefer Chronulus input types such as TextFromFile, PdfFromFile, and ImageFromFile when working with files, and include the forecast explanation supplied by Chronulus when presenting a plot.

How it works

The server operates as a local MCP process. Claude Desktop starts that process using either the Python module, the uvx package runner, or a Docker container. Requests are then handled through the configured Chronulus connection, which requires a Chronulus API key.

The repository’s example configuration registers the server under the name chronulus-agents. Python starts it with python -m chronulus_mcp; the uvx example runs the chronulus-mcp package; and the Docker example passes the API key into the container. These are alternative launch methods, not separate Chronulus services.

Setup and configuration

Install Claude for Desktop first, then add an mcpServers entry to its configuration file. The README identifies the macOS path as ~/Library/Application Support/Claude/claude_desktop_config.json and the Windows path as %APPDATA%\\Claude\\claude_desktop_config.json.

The required environment variable is CHRONULUS_API_KEY. Replace the placeholder in the example configuration with a valid key before starting Claude Desktop. With uvx, the core configuration has this structure:

{
  "mcpServers": {
    "chronulus-agents": {
      "command": "uvx",
      "args": ["chronulus-mcp"],
      "env": {
        "CHRONULUS_API_KEY": "your-key"
      }
    }
  }
}

The package can also be installed from PyPI with pip install chronulus-mcp, or from a cloned repository with pip install .. For Docker, clone the repository, build an image named chronulus-mcp, and run it with standard input enabled while passing CHRONULUS_API_KEY into the container.

Tools and capabilities

ChronulusAI/chronulus-mcp MCP server is intended for conversations involving Chronulus forecasting and prediction agents. Supported workflows described in the README include:

• Sending text, PDF, or image file inputs through the corresponding Chronulus input types.
• Requesting forecasts or predictions from Chronulus agents.
• Using Chronulus-provided forecast explanations when displaying plots.
• Running alongside other MCP servers, such as separately installed filesystem or fetch servers.

The README does not document individual tool names, argument schemas, supported forecast models, or output formats, so those details should be confirmed from the project or Chronulus documentation before building against them.

Limitations and notes

The documented client setup targets Claude for Desktop on macOS and Windows. The examples also depend on external software: Python and the package installation path, uvx, or Docker, depending on the selected method. If Claude reports spawn python ENOENT or spawn uvx ENOENT, the README recommends installing the missing executable or configuring its absolute path.

A Chronulus API key is required, and the provided material does not specify pricing, quotas, regional availability, or API behavior. The filesystem and fetch entries shown in the README are third-party servers used alongside ChronulusAI/chronulus-mcp MCP server; they are not included capabilities of this project. The repository’s license is not stated in the supplied material.