How to add Chronulus AI MCP Server to Claude Desktop
Chat with Chronulus AI forecasting and prediction agents directly from Claude. Paste the config into ~/Library/Application Support/Claude/claude_desktop_config.json and restart Claude Desktop.
Last updated June 14, 2026 · 111★ · stdio · apikey · official
Claude Desktop config for Chronulus AI MCP Server
pip install chronulus-mcp{
"mcpServers": {
"chronulus-ai-mcp-server": {
"command": "uvx",
"args": [
"chronulus-mcp"
],
"env": {
"CHRONULUS_API_KEY": "<YOUR_CHRONULUS_API_KEY>"
}
}
}
}Requires `uv` (the Python package runner). Install it from https://docs.astral.sh/uv/ if `uvx` is not found.
Setup steps
- 1Open Claude Desktop → Settings → Developer → Edit Config (this opens ~/Library/Application Support/Claude/claude_desktop_config.json).
- 2Paste the Chronulus AI MCP Server config below under the top-level "mcpServers" key.
- 3Fill in any placeholder secrets (API keys, paths) in the snippet.
- 4Save the file, then fully quit and reopen Claude Desktop.
- 5Open a chat and confirm Chronulus AI MCP Server's tools appear under the 🔌 tools menu.
Before you start
- A Chronulus API key (CHRONULUS_API_KEY)
- Python 3.10+ (when installed via pip), or uv/uvx, or Docker
- An MCP client such as Claude for Desktop
What Chronulus AI MCP Server can do in Claude Desktop
create_chronulus_sessionCreates a new Chronulus session from a situation and task describing the forecasting/prediction use case and returns a reusable session_id.
create_forecasting_agent_and_get_forecastCreates a NormalizedForecaster agent with a session and input data model, submits forecast input data, and returns prediction data (values between 0 and 1) plus a text explanation.
reuse_forecasting_agent_and_get_forecastSubmits new forecast input to a previously created NormalizedForecaster agent (unchanged input data model) and returns prediction data and a text explanation.
rescale_forecastRescales NormalizedForecaster predictions (values between 0 and 1) to a use-case-specific min/max, with optional inverse scaling.
save_forecastSaves a NormalizedForecaster forecast to separate CSV (data) and TXT (explanation) files, with optional rescaling via y_min/y_max and invert_scale.
create_prediction_agent_and_get_predictionsCreates a BinaryPredictor agent with a session and input data model, submits prediction input, and returns a consensus probability from a panel of experts plus individual estimates, explanations, and Beta-distribution alpha/beta parameters.
reuse_prediction_agent_and_get_predictionSubmits new prediction input to a previously created BinaryPredictor agent (unchanged input data model) and returns the consensus probability, expert estimates, explanations, and Beta-distribution parameters.
save_prediction_analysis_htmlSaves a BinaryPredictor prediction analysis to an HTML file, including a plot of the theoretical and empirical Beta distribution and the experts' opinions.
Security
Requires a Chronulus API key supplied via the CHRONULUS_API_KEY environment variable. Inputs (text and files) are sent to the Chronulus AI cloud platform for processing; total input size cannot exceed 10MB.
Chronulus AI MCP Server + Claude Desktop FAQ
Where is the Claude Desktop config file?
Claude Desktop reads MCP servers from ~/Library/Application Support/Claude/claude_desktop_config.json. Paste the Chronulus AI MCP Server config there under the "mcpServers" key and restart the client.
Is Chronulus AI MCP Server safe to use with Claude Desktop?
Requires a Chronulus API key supplied via the CHRONULUS_API_KEY environment variable. Inputs (text and files) are sent to the Chronulus AI cloud platform for processing; total input size cannot exceed 10MB.
How do I authenticate with Chronulus?
Set the CHRONULUS_API_KEY environment variable in your MCP client config to your Chronulus API key.
What input formats can I pass to the agents?
You can pass plain text or files by path, including images, text files, and PDFs (via input types such as ImageFromFile, TextFromFile, and PdfFromFile). The total size of all inputs cannot exceed 10MB.
Do I need historical data to forecast?
No. Both the NormalizedForecaster and BinaryPredictor agents are designed to produce forecasts and probability estimates without requiring historical data.