AutoTS — Claude Code MCP server: what it does, how to install
Runs automated time series forecasting with genetic model search and probabilistic bounds. Useful when you need quick predictions directly in your assistant.
claude mcp add autots -- uvx autots-mcpWhat AutoTS does
AutoTS searches for optimal forecasting pipelines over pandas.DataFrame inputs. It explores statistical methods, machine learning, and deep learning architectures while applying genetic algorithms to select feature transforms and ensembles. The tool natively handles multivariate series and external regressors.
The server runs via uvx autots-mcp. Default installation includes basic dependencies only, so specialized models require separate Python packages installed in your environment. Input tables work in either wide formats with datetime indexes or long three-column structures.
What it helps with
- find the best forecasting pipeline using genetic search
- generate probabilistic predictions with upper and lower bounds
- build horizontal or mosaic ensembles across multiple series
- forecast multivariate time series with exogenous regressors
Example prompts for the agent
- Forecast values 21 steps ahead from
metrics.csvusing fast models - Fit an ensemble with a 0.9 prediction interval on this wide dataframe
- Find the best time series pipeline for long format data with series identifiers
How to install AutoTS
Platform CLI
claude mcp add autots -- uvx autots-mcpuvx runs the command named after the package; if the README names another one, use uvx --from autots-mcp <command>
Platform CLI
codex mcp add autots -- uvx autots-mcpuvx runs the command named after the package; if the README names another one, use uvx --from autots-mcp <command>
Config file
{
"mcpServers": {
"autots": {
"command": "uvx",
"args": [
"autots-mcp"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from autots-mcp <command>
Config file
{
"servers": {
"autots": {
"type": "stdio",
"command": "uvx",
"args": [
"autots-mcp"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from autots-mcp <command>
Platform CLI
gemini mcp add autots uvx autots-mcpuvx runs the command named after the package; if the README names another one, use uvx --from autots-mcp <command>
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about AutoTS
How to install AutoTS in Claude Code?
Run claude mcp add autots -- uvx autots-mcp in your terminal.
What data structures does AutoTS support?
It processes Pandas DataFrames in wide format with a DatetimeIndex or long format with date, series ID, and value columns.
Are all forecasting models available immediately?
The base package covers standard models. Specific statistical methods and deep learning architectures require extra dependencies.
