HealthTech & Life Sciences No_Free_ContextSubmitted July 26, 2026

NutriPulse - Biometric-Aware Nutrition Intelligence for AI Agents

An MCP app on the Model Context Protocol built by No_Free_Context at the Amrita University Coimbatore NitroStack × MCP To The Moon hackathon and deployed on NitroStack.

About this project

Problem: Ask a Swiggy or Zomato assistant what to eat tonight and it answers from menus, ratings and past orders. It cannot see your HbA1c, your ferritin, or that you slept four hours. It cannot know you're on warfarin, so a spinach curry may interfere with your medication. These platforms optimise for conversion, personalisation stops at cuisine preference. Fitness trackers hold the other half of the picture (sleep, stress, recovery, activity) in dashboards that never reach the ordering decision. Solution: NutriPulse is an MCP server unifying clinical profiles, lab panels, live wearable telemetry and a USDA-resolved food catalogue into one layer any AI agent can reason over. Real-time biometrics drive it. It accepts sleep, stress, HR recovery, hydration, steps and active calories from a smartwatch or band, and targets shift immediately, four hours' sleep with stress at 85 raises protein, lowers the sugar ceiling and increases fluid requirements, each adjustment naming its trigger. Recommendations run a four-axis resolver. Every candidate is scored independently on clinical fit, contextual/taste fit, budget fit and craving satisfaction. The engine computes the Pareto-nondominated set, then applies a lexicographic tiebreak: clinical severity, then budget cap, then craving, then preference. Allergen and drug–nutrient interaction BLOCKs are absolute, enforced by a framework-level guard no tool can bypass. Missing nutrient data fails closed rather than silently passing. Every result returns a conflict log naming what was sacrificed in real units: milligrams, grams, rupees and which rule decided it. Computed, not generated. Impact: India carries one of the world's largest diabetic and hypertensive populations alongside surging delivery adoption. NutriPulse negotiates healthier swaps rather than refusing cravings. External components: NitroStack SDK; USDA FoodData Central (traceable source IDs); Open-Meteo. Synthetic personas; no real patient data.

HealthTech & Life Sciences track

Design AI-powered solutions for healthcare, diagnostics, patient care, medical research, and life sciences.

Source code

Team No_Free_Context

  • M PraneshLead

  • Yadalam Sai Venkata Nihanth

  • Naga Abhinava Sai Juturu

  • Rachuri sai sathvik

← All Amrita University Coimbatore projects

Frequently asked questions

What does NutriPulse - Biometric-Aware Nutrition Intelligence for AI Agents do?
Problem: Ask a Swiggy or Zomato assistant what to eat tonight and it answers from menus, ratings and past orders. It cannot see your HbA1c, your ferritin, or that you slept four hours. It cannot know you're on warfarin, so a spinach curry may interfere with your medication. These platforms optimise for conversion, personalisation stops at cuisine preference. Fitness trackers hold the other half of the picture (sleep, stress, recovery, activity) in dashboards that never reach the ordering decision. Solution: NutriPulse is an MCP server unifying clinical profiles, lab panels, live wearable telemetry and a USDA-resolved food catalogue into one layer any AI agent can reason over. Real-time biometrics drive it. It accepts sleep, stress, HR recovery, hydration, steps and active calories from a smartwatch or band, and targets shift immediately, four hours' sleep with stress at 85 raises protein, lowers the sugar ceiling and increases fluid requirements, each adjustment naming its trigger. Recommendations run a four-axis resolver. Every candidate is scored independently on clinical fit, contextual/taste fit, budget fit and craving satisfaction. The engine computes the Pareto-nondominated set, then applies a lexicographic tiebreak: clinical severity, then budget cap, then craving, then preference. Allergen and drug–nutrient interaction BLOCKs are absolute, enforced by a framework-level guard no tool can bypass. Missing nutrient data fails closed rather than silently passing. Every result returns a conflict log naming what was sacrificed in real units: milligrams, grams, rupees and which rule decided it. Computed, not generated. Impact: India carries one of the world's largest diabetic and hypertensive populations alongside surging delivery adoption. NutriPulse negotiates healthier swaps rather than refusing cravings. External components: NitroStack SDK; USDA FoodData Central (traceable source IDs); Open-Meteo. Synthetic personas; no real patient data.
Who built NutriPulse - Biometric-Aware Nutrition Intelligence for AI Agents?
NutriPulse - Biometric-Aware Nutrition Intelligence for AI Agents was built by team No_Free_Context at the Amrita University Coimbatore NitroStack × MCP To The Moon hackathon, in the HealthTech & Life Sciences track.
What is an MCP app and how is it built?
An MCP app is an application built on the Model Context Protocol — an open standard that lets AI agents connect to tools, data, and APIs. This project exposes MCP tools and resources that agentic AI systems can call. It was built and deployed on NitroStack, the full-stack platform for shipping MCP apps and servers.