When I instantiated AlignerDTW and tried to call mcp_sktime_fit(dataset="airline"), the MCP server returned exactly this:
{
"success": false,
"error": "{'df-list': \"X must be of python type list, or a subtype thereof, but found <class 'pandas.core.series.Series'>\"}"
}Limitation: Aligners mathematically align multiple time series, so their native fit() method strictly requires a Python list of DataFrames/Series (e.g. X=[df1, df2]). The MCP tool mcp_sktime_fit accepts a single dataset parameter and loads a single DataFrame. There is currently no schema design in the tools for an LLM to say "Load these 5 datasets and pass them as a single list to X".
When I instantiated AlignerDTW and tried to call mcp_sktime_fit(dataset="airline"), the MCP server returned exactly this:
{ "success": false, "error": "{'df-list': \"X must be of python type list, or a subtype thereof, but found <class 'pandas.core.series.Series'>\"}" }Limitation: Aligners mathematically align multiple time series, so their native fit() method strictly requires a Python list of DataFrames/Series (e.g. X=[df1, df2]). The MCP tool mcp_sktime_fit accepts a single dataset parameter and loads a single DataFrame. There is currently no schema design in the tools for an LLM to say "Load these 5 datasets and pass them as a single list to X".