from mcp.server.fastmcp import FastMCP
from mcp.server.transport_security import TransportSecuritySettings
import os
import httpx
from dotenv import load_dotenv
from typing import Dict, Any, List
# Import local modules
from src.community.search import HybridSearchClient
from src.treatment.calculator import calculate_bolus
# Load environment variables
load_dotenv()
# Transport security settings for Fly.io deployment
# Allows the Fly.io proxy Host header to pass validation
transport_security = TransportSecuritySettings(
enable_dns_rebinding_protection=True,
allowed_hosts=["t1d-mcp.fly.dev", "t1d-mcp-lite.fly.dev", "localhost", "localhost:8080", "127.0.0.1:8080"],
allowed_origins=["*"], # PlayMCP에서 접근 허용
)
# Initialize MCP Server with transport security
mcp = FastMCP("T1D Manager", transport_security=transport_security)
# Initialize Services
search_client = HybridSearchClient()
@mcp.tool()
def calculate_insulin_dosage(current_bg: int, target_bg: int, isf: int, carbs: int, icr: int) -> str:
"""
Calculate suggested insulin bolus dose for Type 1 diabetes management.
This tool calculates both correction dose (for high blood glucose) and
meal dose (for carbohydrate intake) using standard formulas.
Use this when a user asks:
- "How much insulin should I take?"
- "Calculate my bolus for this meal"
- "I need to correct my high blood sugar"
Args:
current_bg: Current blood glucose level in mg/dL
target_bg: Target blood glucose level in mg/dL (typically 100-120)
isf: Insulin Sensitivity Factor - how much 1 unit lowers glucose (mg/dL)
carbs: Carbohydrates to be consumed in grams
icr: Insulin-to-Carb Ratio - grams of carbs covered by 1 unit
Returns:
JSON with correction_dose, meal_dose, total_dose, and calculation breakdown
"""
import json
result = calculate_bolus(current_bg, target_bg, isf, carbs, icr)
return json.dumps(result, ensure_ascii=False)
@mcp.tool()
def search_diabetes_community(query: str) -> str:
"""
Search Korean blogs and web pages for Type 1 diabetes patient experiences and tips.
Use this tool to find real-world advice from the diabetes community, such as:
- Low blood sugar snack recommendations
- CGM sensor patch tips
- Travel advice for diabetics
- Daily life management tips
Args:
query: Search query in Korean or English
Returns:
JSON array of search results with title, link, and source
"""
import json
results = search_client.search_hybrid(query)
# Add source icons for better structured data context
for item in results:
if item.get("source") == "Naver Blog":
item["icon"] = "🟢"
elif item.get("source") == "Daum Web":
item["icon"] = "🟡"
return json.dumps(results, ensure_ascii=False)
@mcp.tool()
def analyze_sick_day_guidelines(
symptoms: str = "감기 기운",
current_glucose_mg_dl: int | None = None,
ketone_mmol_l: float | None = None
) -> str:
"""
Analyze sick day risks and retrieve clinical guidelines based on patient status.
Use this tool when the user reports being unwell. It returns evidence-based
recommendations (JSON) from ISPAD/ADA guidelines to help the LLM generate
safe and personalized medical advice.
Args:
symptoms: User's reported symptoms (comma-separated).
current_glucose_mg_dl: Current blood glucose in mg/dL (optional).
ketone_mmol_l: Current blood ketone level in mmol/L (optional).
Returns:
JSON string containing risk level, advice, and emergency warnings.
"""
import json
from src.treatment.sick_day.risk_analyzer import (
analyze_sick_day_risk,
serialize_sick_day_result
)
result = analyze_sick_day_risk(
symptoms_input=symptoms,
glucose_mg_dl=current_glucose_mg_dl,
ketone_mmol_l=ketone_mmol_l
)
# Return JSON string for MCP compatibility
return json.dumps(serialize_sick_day_result(result, symptoms), ensure_ascii=False)
# ===== Dexcom Developer API (Official Sandbox) Tools =====
# Conditionally enabled via ENABLE_DEXCOM environment variable
# These tools require OAuth authentication
ENABLE_DEXCOM = os.getenv("ENABLE_DEXCOM", "true").lower() == "true"
if ENABLE_DEXCOM:
from src.tools.dexcom_tools import register_dexcom_tools
register_dexcom_tools(mcp)
