"""Aggregation strategy justification prompt."""
from __future__ import annotations
from fastmcp import FastMCP
from fastmcp.prompts import Message, PromptMessage
def register(mcp: FastMCP) -> None:
"""Register aggregation justification prompt."""
@mcp.prompt(
name="justify_aggregation_strategy",
description="Advisory guidance for aggregation/statistical reasoning.",
tags={"reasoning", "aggregation"},
)
def justify_aggregation_strategy(
aggregation_type: str,
statistic: str,
zones: str | None = None,
operation_context: str | None = None,
) -> list[PromptMessage]:
extras = []
if zones:
extras.append(f"Zones: {zones}")
if operation_context:
extras.append(operation_context)
context_suffix = f" ({', '.join(extras)})" if extras else ""
content = (
f"The operation will use **{statistic}** statistic for **{aggregation_type}**"
f"{context_suffix}.\n\n"
"**Your role:**\n"
"• If the user specified this statistic: document why it's appropriate. "
"If you see concerns, ask them conversationally before proceeding.\n"
"• If you're choosing this statistic: explain your reasoning.\n\n"
"**Consider:**\n"
"• Interpretation goal? (central tendency/dominance/extremes)\n"
"• How does statistic preserve phenomenon?\n"
"• Distribution/independence/zone design assumptions?\n"
"• Why not mean/median/mode/sum/min/max?\n\n"
"**Return strict JSON:**\n"
"```json\n"
"{\n"
' "intent": "interpretation goal (e.g., typical value, dominant class)",\n'
' "alternatives": [\n'
' {"method": "mean|median|mode|sum|min|max", "why_not": "reason"}\n'
" ],\n"
' "choice": {\n'
f' "method": "{statistic}",\n'
' "rationale": "why this statistic captures the intent",\n'
' "tradeoffs": "information lost or de-emphasized"\n'
" },\n"
' "confidence": "low|medium|high"\n'
"}\n"
"```"
)
return [Message(content=content, role="user")]
