MCP projects built at Amrita University Amritapuri campus
These are 176 real MCP apps and servers built and deployed by student teams at the Amrita University Amritapuri campus NitroStack × MCP To The Moon hackathon. Each project is a working agentic AI build on the Model Context Protocol, shipped on NitroStack. Open any project to watch the demo, read the write-up, and explore the source.
What does it do?
The Multi-Domain Semantic Architect Agent is an enterprise-grade Model Context Protocol (MCP) server that acts as a cognitive bridge between local software development, global research registries, and aesthetic design moodboards. Spanning three distinct domains, the agent aggregates local workspace AST parsing, academic preprint cross-pollination, and visual UI synthesis into a single unified workflow. It autonomously audits framework vitality, forecasts dependency conflicts, scans for CVE vulnerabilities, and physically scaffolds secure boilerplate code and Docker containers directly into the IDE.
Who is it for?
This tool is built for developer teams, software architects, and AI coding assistants (such as Claude Desktop or Cursor). It empowers developers to design secure, compliant, and highly optimized software architectures while eliminating decision fatigue and the fear of long-term technical debt.
What makes it special?
Unlike standard code generators, this is a fully autonomous executor deeply integrated with the Nitrostack ecosystem. It boasts 24 distinct autonomous analysis tools and leverages Nitrostack's Zod-based middleware for "Autonomous Schema Auto-Healing"—instantly intercepting and correcting LLM hallucinations before they break the application. It seamlessly translates complex mathematical formulas from arXiv into actionable code, enforces strict supply-chain security gates, and synthesizes frontend UI palettes. Finally, instead of returning walls of text, it presents its data through beautiful, interactive React components (like the Framework Health Dashboard) rendered natively via @nitrostack/widgets.
Here's a shorter version (around **110–130 words**):
> **SmartClaim** is an AI-powered insurance claim fraud detection system built using **NitroStack MCP**, **TypeScript**, and **Google Gemini AI**. It analyzes accident details, vehicle damage images, repair bills, customer claim history, weather conditions, and location data to identify potentially fraudulent insurance claims. Gemini AI estimates vehicle damage severity, repair costs, and extracts information from repair bills, while other services verify accident details and customer history. Based on all collected evidence, the system generates a fraud score, confidence level, risk category, and recommendations for investigators. SmartClaim aims to reduce manual verification efforts, improve claim processing speed, and enhance fraud detection accuracy by combining AI-powered image analysis with intelligent rule-based validation.
Aasha connects India's fragmented health systems like maternal health, vaccination records, nutrition data, digital IDs, and vaccine stock into a single AI-accessible platform. Instead of healthcare workers jumping between disconnected government portals to find patient information, they can ask a simple question like "Show me unvaccinated children at high risk" and get instant answers with visual insights.
The platform is built for healthcare workers, and district administrators managing India's public health system,anyone who's struggled with manual cross-referencing across multiple systems. Aasha automatically queries multiple government registries in the background, connects the dots, and delivers integrated intelligence without the administrative overhead.
What makes Aasha special is its combination of speed and security. Healthcare workers get instant answers with interactive dashboards for visualization, while real governance keeps data safe and prevents raw information from being exposed. Pre-loaded with production-ready synthetic data, it works immediately. Aasha turns siloed, fragmented data into actionable intelligence that saves time and improves patient outcomes
Proactive Data Leak Containment Agent is an MCP-powered security assistant that continuously monitors developer files for exposed credentials. When a secret is detected, the agent automatically quarantines the affected file, updates the MCP security state, logs the incident, generates a detailed forensic report, and displays live security statistics. This demonstrates autonomous threat detection, containment, and incident reporting using the Model Context Protocol.
Campus Pilot is an MCP-based AI agent that acts as a campus operations copilot — checking schedules, room availability, and events, then actually acting on them (booking rooms, logging issues). Its standout feature is a real hardware tool integration: one demo room wired with a sensor and relay, so the agent can read live occupancy/conditions and physically toggle a light or fan. This grounds the "agentic AI" pitch in something judges can see happen in real time, not just database lookups. It reuses your embedded systems background as a genuine technical differentiator. The result: a practical, demoable tool rather than another generic chatbot.
MedForge is an AI-powered Smart Hospital Assistant built using the *NitroStack MCP Framework*. It showcases a Model Context Protocol (MCP) first architecture, combining type-safe tools, resources, system prompts, and responsive visual widgets to guide patients from initial symptom triage to room-by-room hospital navigation.
The reviewer-zero project is an advanced, automated academic peer-review system built on the NitroStack framework. Its primary objective is to streamline the peer-review process by utilizing a multi-agent architecture where specialized AI tools handle distinct aspects of paper analysis, evaluation, and synthesis.
At the core of the backend architecture are three main components:
ReviewerModule: This houses a suite of 16+ specialized AI agents designed for intricate peer-review tasks. Key agents include the ClaimsAuditorTool (for verifying logical soundness), RigorAgentTool (for assessing methodology), CitationScoutTool (for checking reference validity), and WeaknessDetectorTool. These agents are powered by an LLMService (integrating with models via Groq/Gemini), a SemanticScholarService for academic lookups, and a LatexParserService for processing mathematical formatting.
Aegis Interceptor: To ensure stability and resilience, the AegisInterceptor acts as a self-healing middleware during tool executions. It catches failures—such as API rate limits or network errors from LLM providers—and automatically attempts retries using exponential backoff. It meticulously logs all these healing events so users can monitor system resilience.
OrchestratorModule: This module acts as the "manager" of the system. Its OrchestratorTool coordinates the workflow of the various review agents, taking a paper as input, delegating tasks, and synthesizing their independent findings into a cohesive, structured ReviewPayload. This ultimately produces a comprehensive ReviewLetter and a final grading of the paper.
On the frontend, the project features a modern, interactive web dashboard built with Next.js (src/widgets/app/review-dashboard). This dashboard serves as the user-facing interface, allowing researchers to visualize the real-time execution of the AI agents, monitor system health (such as Aegis healing events), and read the synthesized peer reviews. The UI is designed with a premium
An AI-powered FinOps assistant that analyzes live AWS cloud resources, detects idle and underutilized EC2 instances using real CloudWatch metrics, estimates potential cost savings, and provides actionable optimization recommendations through NitroStack's Model Context Protocol (MCP).
Sietch - Policy Red-Flag & Bill Auditor is an AI-powered healthcare insurance assistant built as an MCP (Model Context Protocol) server using NitroStack. The platform simplifies the insurance claim process by helping users understand complex policy documents, estimate claim payouts, locate network hospitals, and analyze hospital bills before submitting a claim.
The system consists of four intelligent tools:
Policy Analyzer – Reads insurance policy documents, identifies exclusions, waiting periods, sub-limits, co-pay clauses, and other hidden conditions, while explaining them in simple language.
Coverage Calculator – Calculates the estimated insurance payout for a specific diagnosis by considering policy rules such as waiting periods, sub-limits, co-pay percentages, and coverage eligibility.
Network Hospital Lookup – Finds nearby hospitals that are covered under the user's insurance policy, helping users identify cashless treatment facilities quickly during emergencies.
Bill Auditor – Examines hospital bills item by item, estimates what the insurance is likely to cover, highlights non-covered expenses, detects possible overcharges, and provides a plain-language explanation of every charge to help users avoid unnecessary costs.
Built with TypeScript, NitroStack, Groq LLM, and Zod schemas, the solution processes documents securely in memory without persistent storage, ensuring user privacy. As an MCP server, it can be integrated into hospitals, insurance companies, customer support systems, or AI assistants, enabling faster, more transparent, and more reliable insurance claim decisions.
ARTHA AI is an autonomous multi-agent financial intelligence platform designed to democratize professional-grade investment advisory. It combines specialized AI agents for market analysis, risk profiling, portfolio optimization, financial news intelligence, and personalized wealth planning into a single conversational experience.
Unlike traditional investment platforms that provide static dashboards or generic recommendations, ARTHA AI continuously analyzes real-time market signals, user risk appetite, investment goals, and financial trends to generate transparent, explainable, and personalized investment strategies.
Built using an AI-native architecture with Model Context Protocol (MCP), NitroStack, Supabase, and modern web technologies, the platform enables intelligent decision-making through modular AI tools, interactive widgets, and autonomous workflows.
ARTHA AI empowers students, retail investors, working professionals, and first-time investors with institutional-quality financial insights—making wealth creation more accessible, transparent, and data-driven.
Anvaya is an AI Business Operating System for small manufacturers, wholesalers, farmers, and retailers. Instead of just answering questions, it acts on its own — finding the most profitable place to sell, discovering the best buyers or suppliers, comparing transport to cut costs, suggesting negotiation terms, and automatically generating invoices while updating inventory.
It's built for MSMEs who currently run on guesswork and manual coordination across sales, buying, logistics, and paperwork. Every recommendation comes with a clear "why," so it feels less like a tool and more like an AI-powered business partner running things end-to-end.
MyCarCompanion is an AI powered MCP server that helps car owners monitor car health, fuel and maintenance through natural language interactions. Built using NitroStack and SmartCar APIs, it goes beyond traditional vehicle apps by providing intelligent, personalised recommendations instead of just displaying data.
Modern cloud operations require engineers to constantly switch between dashboards for monitoring, billing, policy compliance, security, and infrastructure management. This fragmented workflow increases operational complexity and slows down decision-making.
NimbusOps is an Agentic Cloud Operations Copilot built with NitroStack and the Model Context Protocol (MCP) that transforms natural language requests into intelligent cloud workflows. Instead of interacting with multiple cloud services manually, users simply describe their objective, and NimbusOps automatically plans, orchestrates, and executes the appropriate MCP tools to generate a unified operational response.The system follows a modular, enterprise-ready architecture where each cloud capability—such as billing analysis, rightsizing recommendations, policy evaluation, rollback simulation, and infrastructure insights—is implemented as an independent MCP tool. An AI planning engine performs intent classification, selects the required tools, invokes them dynamically, and aggregates their structured outputs into a single actionable report.
Built using a TypeScript-based NitroStack MCP Server and a FastAPI backend, NimbusOps demonstrates how agentic AI, tool orchestration, and the Model Context Protocol can simplify cloud operations while improving scalability, extensibility, and automation.
Key Features
* 🤖 Agentic AI Planning Engine
* 🔗 Dynamic MCP Tool Orchestration
* ☁️ Cloud Billing & Cost Analysis
* 📈 Rightsizing Recommendations
* 📋 Policy Compliance Evaluation
* 🔄 Deployment Rollback Simulation
* 💬 Natural Language Infrastructure Queries
* 🏗️ Modular, Extensible Enterprise Architecture
* ⚡ Built with NitroStack MCP Framework & FastAPI
HomeOps AI is an Agentic AI assistant built using the NitroStack SDK and the Model Context Protocol (MCP). Instead of relying on a single chatbot, the assistant intelligently understands user intent and dynamically invokes specialized MCP tools such as diagnosis, maintenance, repair guidance, safety checks, technician assistance, shopping recommendations, planning, and long-term memory. The system provides personalized and context-aware assistance for everyday home operations through natural language conversations.
Space is entering its busiest era. With thousands of satellites already in orbit and tens of thousands more on the way, collision avoidance can no longer depend on fragmented communication and manual human decisions. A single mistake can create thousands of pieces of debris, triggering cascading collisions that threaten critical infrastructure, global communications, navigation, weather forecasting, and future space missions.
VIYAN reimagines how satellites coordinate in space.
Instead of simply detecting potential collisions, VIYAN enables autonomous, intelligent decision-making. Using real-time orbital data, our platform predicts future conjunctions, evaluates risk through a hybrid AI engine combining orbital physics with machine learning, and enables AI agents representing different satellite operators to negotiate the safest course of action.
Unlike traditional systems that only generate warnings, VIYAN answers the critical question: "What should happen next?"
Our platform performs the complete decision pipeline:
Predicts future satellite trajectories.
Detects potential conjunctions before they become dangerous.
Evaluates collision probability using a hybrid physics + AI model.
Negotiates between satellites based on mission priority and operational constraints.
Recommends the safest maneuver with confidence scores.
Explains every decision using Explainable AI (SHAP), ensuring transparency and trust.
The result is a platform that transforms collision management from a reactive alerting system into an autonomous coordination ecosystem.
Key Innovations
🛰️ Real-time orbital trajectory prediction.
🤖 Hybrid AI risk assessment combining deterministic physics and machine learning.
🤝 Autonomous multi-agent negotiation between satellite operators.
🧠 Explainable AI with transparent reasoning behind every recommendation.
📊 End-to-end decision support with timelines, confidence scores, and maneuver suggestions.
ADStrike is an AI-powered MCP server for Active Directory security assessments that performs live LDAP enumeration, attack path analysis, and MITRE ATT&CK mapping.
It automatically discovers AD assets, identifies privilege escalation opportunities, validates attack paths, and generates prioritized remediation reports.
Designed for authorized red team engagements, ADStrike helps security teams visualize domain risk and strengthen Active Directory defenses.
PathFinder AI is an intelligent learning assistant built using NitroStack MCP (Model Context Protocol). It helps students discover structured learning resources, understand prerequisite concepts, follow personalized learning roadmaps, and receive curated book recommendations—all through an interactive AI-powered interface.
Most drug-interaction tools are built on US drug databases and are effectively blind to Indian medicine. Ask any of them about "Dolo 650" or "Combiflam" and they return nothing — because international drug registries like RxNorm don't index Indian brand names. Yet these are exactly the medicines an Indian patient actually holds in their hand. MedSafe India closes that gap.
It's an MCP server that lets any AI assistant check a patient's medications for two real dangers: high-priority drug interactions, and — the more common, more invisible risk — duplicate active ingredients across combination products. A patient taking Dolo 650 and Combiflam for fever is taking paracetamol twice without knowing it, a genuine route to liver injury. MedSafe catches this in one step.
🚀 NexusIQ MCP
🤖 AI-Powered Procurement Decision Agent
NexusIQ is an MCP-native Agentic AI that helps organizations answer one simple but critical question:
"Should I sign this deal?"
Instead of following a fixed workflow, NexusIQ reasons like an experienced procurement analyst, deciding what information it needs and dynamically selecting the appropriate MCP capabilities before making a recommendation.
🧠 How It Works
📄 Upload a Procurement Contract
⬇️
🤖 AI analyzes the contract and reasons about potential risks
⬇️
🛠️ The agent dynamically decides which MCP capabilities to use
📑 Contract Analyzer
🏢 Vendor Intelligence
💰 Cost Impact Simulator
(The agent may use all, some, or none depending on the contract.)
⬇️
📚 Retrieves Vendor Knowledge through MCP Resources
⬇️
💡 Generates negotiation strategies using MCP Prompts
⬇️
✅ Provides a final procurement recommendation
✔️ Sign
✏️ Renegotiate
❌ Don't Sign
Along with:
📌 Key contract clauses
🏢 Vendor insights
💵 Financial impact estimates
🤝 Negotiation recommendations
⚡ Why NexusIQ?
Unlike traditional document-analysis tools, NexusIQ doesn't execute a predefined pipeline.
It reasons before acting, choosing the right MCP Tools, Resources, and Prompts based on the contract's content—making every analysis adaptive, context-aware, and genuinely agentic.
🛠️ MCP Components
🔧 Tools
📑 Contract Analyzer
🏢 Vendor Intelligence
💰 Cost Impact Simulator
📚 Resource
🗂️ Vendor Knowledge Base
💬 Prompt
🤝 Negotiation Assistant
🎯 Goal
Demonstrate how Model Context Protocol (MCP) enables AI agents to intelligently interact with external tools and knowledge sources, helping enterprises make faster, smarter, and more confident procurement decisions.
I actually like ending it with a short tagline:
🧠 Think. Analyze. Negotiate. Decide.
NexusIQ transforms procurement from a checklist into an intelligent conversation.
EduAccess AI is an intelligent learning platform that bridges the language gap in education. It automatically translates teachers' content into students' local languages, provides personalized learning support, and uses Agentic AI with MCP to make quality education accessible for rural students.
AI-powered recovery agent that helps patients transition safely from hospital to home. Instead of leaving patients with complex discharge summaries filled with medical jargon, the agent converts them into a clear, personalized recovery plan that is easy to understand and follow.
Our project is submitted under the Open Innovation track because agricultural resource management represents a unique intersection of biology, environmental science, and hyper-local data that does not fit neatly into the other five tracks. While it utilizes sensor data (similar to IoT), the end goal is not industrial automation, but providing actionable, biological guidance to farmers regarding crop health, soil pH balancing, and weather risk mitigation.
CampusOS AI is an AI-powered University Operating System built using Model Context Protocol (MCP) and Agentic AI. It unifies essential campus services such as attendance, timetables, assignments, library management, events, fees, and academic analytics into a single intelligent platform. Instead of navigating multiple university portals, students, faculty, and administrators interact with one AI assistant that can answer questions, automate tasks, and provide personalized insights. CampusOS AI leverages modular MCP servers to connect university systems, creating a smarter, more efficient, and connected digital campus experience.
InvestIQ is an autonomous investment analyst built entirely on orchestrated MCP tools — not a chatbot wrapper around one API call. Ask "Analyze NVIDIA" (or any global stock) and the host LLM discovers and chains 7 independent tools: get_stock_quote → get_company_profile → get_basic_financials → get_company_news → summarize_news → analyze_stock → generate_investment_report, each rendering its own dark-mode widget with a live execution timeline that makes the orchestration visible to the user.
Who it's for: retail investors, finance students, and anyone who wants a fast, explainable read on a stock without a paid terminal.
What makes it special:
• Real multi-tool orchestration, not a monolithic "analyze" endpoint — each tool is independently reusable
• Deterministic, rule-based analysis engine — every strength/weakness/risk cites the actual number behind it (e.g. "P/E of 54.2 is well above the ~20x market average"), not an LLM guess. No financial advice, full reasoning transparency.
• True global coverage: Finnhub for US markets, with an automatic Yahoo Finance fallback for exchanges Finnhub's free tier can't serve (NSE/BSE India, etc.) — including plain-English name resolution, so "analyze reliance" or "analyze itc" just works
• Every response is tagged with its actual data source (Finnhub vs Yahoo Finance) for full transparency
• Zero conversational logic in the server itself — it's pure, composable MCP tools; all orchestration and synthesis happens at the host LLM layer
Veritas AI is a multi-agent crisis verification system built on NitroStack MCP that helps detect misinformation and verify public claims. When a user submits a claim, specialized AI agents independently analyze news sources, government data, supporting evidence, event timelines, and source credibility. A Decision Agent then combines these analyses to generate a final verdict, confidence score, risk level, and recommendation. The results are presented through an interactive NitroStack widget, enabling users to quickly assess the authenticity of crisis-related information in a clear and user-friendly way.
Compliance Fabric MCP is an enterprise-grade Model Context Protocol (MCP) platform that transforms regulatory compliance into reusable AI infrastructure. Instead of embedding KYC, AML, document verification, and audit logic into every application, organizations integrate once with Compliance Fabric MCP and expose standardized compliance tools that AI agents, workflows, and enterprise applications can invoke.
The platform centralizes identity verification, document intelligence, KYC validation, AML screening, sanctions and PEP checks, risk scoring, policy validation, explainable AI decisions, and immutable audit logging. It provides reusable MCP tools such as verify_identity(), verify_document(), aml_screen(), sanctions_check(), risk_score(), policy_validate(), and audit_generate() to automate end-to-end compliance workflows.
Designed for banks, FinTechs, insurers, and other regulated industries, Compliance Fabric MCP integrates with enterprise systems and trusted data sources while ensuring transparent, consistent, and regulator-ready decisions. By separating compliance from application logic, it reduces duplicate implementations, accelerates onboarding, lowers engineering effort, simplifies audits, and enables organizations to build secure, scalable, and compliant AI-powered applications faster
HandoffOS is an MCP-native workflow state engine that makes invisible enterprise handoffs visible, explainable, and actionable. It transforms workflow events into a live dependency graph, detects bottlenecks with deterministic rules, and uses AI only to explain findings with evidence. Through MCP Resources, Tools, and Prompts, any compatible client can inspect workflow state, simulate the impact of resolving blockers, and execute approved next actions. Our prototype demonstrates this using a new-hire onboarding workflow where a single missed task can delay multiple downstream processes.
DrugWise is an MCP (Model Context Protocol) server that provides AI-assisted clinical decision support for medication safety. It integrates trusted biomedical sources, such as OpenFDA, PubMed, and ClinicalTrials.gov, to provide comprehensive drug information, detect potential drug interactions, retrieve supporting clinical evidence, and suggest possible therapeutic alternatives. Rather than making prescribing decisions, DrugWise empowers healthcare professionals with consolidated, evidence-based insights to support safer and more informed treatment decisions.
CreditPilot is an AI-powered MSME credit intelligence platform that automates credit assessment, financial health analysis, and loan eligibility evaluation using AI and document intelligence.
AML Risk Analyzer is an AI-powered compliance assistant built with NitroStack and the Model Context Protocol (MCP) to streamline Anti-Money Laundering investigations. Using a single intelligent workflow, it performs sanctions screening, adverse media analysis, AML risk assessment, and automatically generates an Enhanced Due Diligence (EDD) report. The solution demonstrates how Agentic AI can orchestrate multiple compliance tools to reduce manual effort, improve investigation consistency, and accelerate decision-making. Designed with a modular MCP architecture and an interactive frontend, the copilot showcases a scalable approach to modern financial compliance, helping institutions identify high-risk customers faster while providing transparent, structured, and actionable investigation reports.
Every minute is critical during organ transplantation. While medical compatibility testing remains essential, identifying suitable donors, verifying consent, and coordinating with hospitals often consumes valuable time.
LifeLink AI is an AI-assisted organ donor intelligence platform designed to streamline this process by creating a centralized donor profile system. Individuals can voluntarily register as organ donors, securely store their medical records, maintain an AI-generated health profile, and digitally manage their donation consent.
When a registered donor is legally declared deceased by a verified medical institution, the platform activates the donor profile for matching. Hospitals and recipient coordinators can submit recipient details such as required organ, blood group, urgency level, and medical constraints. LifeLink AI then intelligently filters potential donors based on consent status, organ availability, blood group compatibility, medical suitability, and geographical proximity before presenting ranked recommendations with transparent reasoning.
The platform also provides AI-generated medical summaries, compatibility scoring, hospital recommendations, and a complete audit trail to support faster and more informed transplant coordination.
This project demonstrates how modern AI, structured medical data, and intelligent workflow automation can significantly reduce donor identification time while supporting healthcare professionals with explainable, decision-support tools. LifeLink AI is intended to assist medical teams—not replace clinical judgement or mandatory medical evaluation.
Factory OS – Problem Statement & Solution
Project Overview
Factory OS is an AI-powered manufacturing intelligence platform that automates quality inspection, procurement, and supply chain management using four specialized AI agents. It enables manufacturers to detect defects in real time, make instant decisions, automatically procure replacement parts, and monitor factory operations through a unified dashboard.
Problem Statement
Manufacturing industries face three major challenges that reduce efficiency and increase operational costs.
1. Slow Quality Control:
Most factories still depend on manual visual inspection to identify defects. This process is slow, inconsistent, and prone to human error, allowing defective products to reach customers and increasing rework, recalls, and financial losses.
2. Inefficient Procurement:
When a defective part is detected, procurement teams manually search for suppliers, compare prices, and place purchase orders. This process often takes hours or days, resulting in production delays, downtime, and increased operational costs.
3. Lack of Real-Time Visibility:
Factory managers often rely on historical reports rather than live production data. Without real-time monitoring and automated decision-making, issues are identified too late, leading to reduced productivity and slower response times.
Our Solution
Factory OS solves these problems using four AI agents that work together in real time.
AI Vision Classifier
The Vision Classifier continuously monitors products on the production line using computer vision. It detects structural defects and surface anomalies with 98.2% accuracy in just 45 milliseconds, preventing defective products from moving further in the manufacturing process.
Triage Agent
Once a defect is detected, the Triage Agent instantly decides whether the product should be accepted, repaired, or rejected. It automatically routes the product to the appropriate workflow within 12 milliseconds, eliminating manual d
Sentinel AI
What it does:
Sentinel AI is an AI-powered cybersecurity assistant that helps users detect, analyze, and respond to cyber threats through a conversational interface. It can identify phishing emails, scan suspicious URLs, retrieve CVE vulnerability information, and generate executive incident reports—all from simple natural language prompts.
What makes it special:
Unlike traditional security tools that require switching between multiple platforms, Sentinel AI brings multiple cybersecurity capabilities into a single AI-powered workspace. It automates threat analysis, correlates information from different security sources, and presents clear, actionable insights, making cybersecurity investigations faster and easier.
Who it's for:
Sentinel AI is designed for cybersecurity professionals, SOC analysts, IT administrators, students, researchers, and organizations looking for a simple yet powerful way to investigate threats, assess vulnerabilities, and improve incident response without requiring extensive manual analysis.
Built StockerMCP – MCP server for market data, paper trading & charting
Tired of finance MCP servers dying the second one API rate-limits you, so I built FinMCP Hub — routes requests to whichever provider fits best, with automatic failover.
Highlights:
Hybrid routing: Twelve Data (quotes/OHLC), Finnhub (profiles/news), Alpha Vantage (statements/ratios), Yahoo Finance India (.NS/.BOM)
Failover: auto-switches providers if one's down/rate-limited; falls back to cache only as last resort; critical ops like portfolio valuation can disable fallback to avoid stale prices
Paper trading: multi-currency (INR + USD), separate portfolios, token-based order preview/confirm with expiry
Charts: Node spawns a Python sidecar (mplfinance) for candlestick + volume charts, returned as Base64 PNGs
Analytics: latency/status logging on every call for easy debugging
Dashboard: Overview / Financials / News / Chart tabs, with validation to catch malformed responses
FinGate MCP: Enterprise Compliance Gateway
With EU AI Act enforcement looming, financial institutions face an infrastructure crisis. Standard MCP servers act as unmonitored open pipes between AI and banking APIs, lacking the human oversight mandated for High-Risk Systems. FinGate MCP is a proxy layer between orchestration clients and downstream tools, enforcing strict structural validation, security, and human mediation.
What It Does & Who It Is For
It transforms AI ecosystems into compliant architectures by intercepting high-risk operations, forcing explicit authorization, and logging actions. Built for BFSI Compliance Officers ensuring regulatory adherence, Security Engineers blocking prompt injections, and AI developers deploying workflows.
Four Security Pillars
🛡️ HITL Mediation: High-risk operations (e.g., $10K transfers) trigger an interception gate. The system suspends execution and routes an approval prompt via /api/mediate before edge-native rehydration.
🧠 Threshold Tuning: Using Bayesian Optimization, FinGate dynamically tunes risk thresholds based on historical approval patterns, reducing false positives while keeping strict 95th-percentile safety limits.
🔒 Cryptographic Verification: To block Man-in-the-Middle exploits, FinGate uses SHA-256 intent hashing. Human-approved parameters are cryptographically mapped directly to the backend payload.
📜 Audit Logging: Every intervention and user action is immutably logged via a SQLite Prisma model (MediationLog) for non-repudiable regulatory compliance.
Domain-Driven Architecture
Host App (Next.js): Handles stream routing (src/). Domains: agent/ (AI logic), security/ (HITL/Bayesian engine), infrastructure/ (Prisma bounds), and interfaces/ (TS definitions).
MCP Server (NitroStack): The secured endpoint (mcp-server/) containing banking tools and strict Zod validations.
Run bun run scripts/pentest.ts to test adversarial payloads and cryptographically prove intercept compliance.
Every important decision a company makes is discussed across multiple platforms like Slack, Notion, emails, GitHub, and calendars. Over time, this information gets scattered, making it difficult to remember why a decision was made, who approved it, and what happened after it was implemented.
SentinelX is an AI-powered enterprise cybersecurity platform built using NitroStudio and deployed on NitroStack. It takes use of MCP to provide intelligent cybersecurity tools that automate critical security operations such as log analysis, IP and URL reputation checks, CVE vulnerability lookups, and incident report generation. Designed for security analysts, SOC teams, and enterprise IT departments, SentinelX streamlines threat detection and incident response through a unified AI-driven workflow. By exposing these capabilities as MCP tools, the platform enables seamless integration with AI agents and modern enterprise systems, reducing manual effort, accelerating investigations, and improving overall cybersecurity operations.
optimized agentic orchestration for medical emergencies. connecting unstructured requests to secure, open-standard knowledge structure via the Model Context Protocol
NexusAI – AI Family Guardian is an intelligent multi-agent platform that helps families securely manage their everyday responsibilities through AI. Built using the Model Context Protocol (MCP), NexusAI orchestrates specialized AI agents for document management, healthcare, government services, financial planning, education, and life-event planning.
Instead of navigating multiple websites and applications, users interact with a single AI assistant that understands their needs and delegates tasks to specialized MCP servers. These agents work together to organize important documents, identify eligible government schemes, summarize medical records, assist with financial decisions, and generate personalized action plans for major life events.
By combining multiple AI agents into one collaborative platform, NexusAI transforms fragmented family services into a unified, secure, and intelligent experience. Our goal is to reduce complexity, save time, and help families make informed decisions with confidence.
Kaigo - MCP-powered agentic AI financial companion designed to simplify and secure financial management for elderly individuals. It enables users to manage pensions, insurance policies, banking services, bill payments, and digital transactions through natural voice or chat interactions. Kaigo proactively tracks pension credits, reminds users of insurance premium due dates, assists with policy renewals, and helps manage everyday financial tasks. It also acts as an intelligent financial guardian by detecting phishing attempts, fraudulent calls, fake payment requests, and banking scams before they can cause harm. By combining agentic AI with secure tool integration, Kaigo delivers a personalized, accessible, and trustworthy financial assistant that empowers senior citizens to confidently navigate modern banking and financial services while providing peace of mind to their families.
The objective of this project is to develop an AI-powered system that can automatically read and interpret medical prescriptions, generate personalized medication schedules, provide timely reminders, monitor medicine inventory, detect potential drug interactions and allergies, recommend equivalent medicine alternatives, and securely manage patient medical history.
The platform will also maintain medicine inventory, notify patients before medicines run out via Telegram and Gmail, set medication alarms, identify drug interactions and allergy risks, recommend medicines with the same active composition along with price comparisons, and allow patients to securely share their medical history with authorized healthcare providers.
WealVest AI demonstrates how the Model Context Protocol (MCP) can power modular AI-driven financial tools for investment analysis, risk assessment, market intelligence, and report generation.
LogGuardian AI is an AI-powered enterprise log analysis platform built with NitroStack MCP and Supabase. It uses a multi-agent architecture to analyze uploaded logs, detect anomalies, identify root causes, generate intelligent recommendations, and produce incident reports through an interactive SOC dashboard.
Verdict is an agentic MCP application that stress-tests business proposals before they reach a real board. Submit any plan and five adversarial AI personas — a Skeptical Investor, Regulator, Competitor, Disgruntled Customer, and Future Historian — independently analyze it using real MCP tools, argue with each other in a live transcript, and deliver a verdict that preserves unresolved disagreement instead of a single smoothed-over answer.
IT IS A RESEARCH TOOL INTEGRATED WITH AN MCP WHICH IS ALSO A LATEX EDITOR .THIS IS JUST A PROOF OF CONCEPT OF A COMPLEX PRODUCT WHICH IS SET TO A SMALLER PROTOTYPE WHICH ISNT EVEN 10 PERCENT OF THE PROJECT . THE SCALABILITY OF THE PRODUCT IN LARGE CUZ THERE IS NO DIPRESSION IN THE GRAPH OF RESEARCHERS AND RESEARCH INTRESTED PEOPLE IN THE LAST DECADE. THE TOOLS USE SOME MAJOR MCP TECHNIQUES TO BUILD THE PROTOTYPE OF THE LARGER CONCEPT . THE IMPACT OF THE TOOL COULD REDUCE TIME ,MONEY AND EFFORT WHICH ARE THE MOST IMPORTANT FACTORS.
FarmAssist AI is an AI-powered crop health assistant that helps farmers identify crop diseases using crop type, symptoms, images, location, and optional soil data. It provides disease diagnosis, confidence scores, treatment recommendations, prevention strategies, and environmental insights through an intuitive interface, enabling faster and more informed farming decisions.
QuickTDS is an AI-ready TDS workflow automation MCP server built to transform how enterprises manage TDS credit reconciliation and recovery. Instead of finance teams manually navigating thousands of invoices, payments, bank transactions, and tax records, QuickTDS connects these fragmented data sources into one automated workflow.
It intelligently links transactions, calculates expected TDS, reconciles deductions against Form 26AS, and detects missing credits, deduction shortfalls, section mismatches, and potential PAN-related discrepancies. When an issue is identified, QuickTDS can turn it into an actionable recovery case, enabling finance teams to move from simply finding discrepancies to actually recovering lost tax credits.
The result is a faster, scalable, and transparent TDS workflow that helps enterprises reduce manual reconciliation effort, identify discrepancies early, and recover money that could otherwise remain unnoticed.
SPECTRA AI is an Agentic AI-powered banking operations platform designed to automate the investigation and resolution of payment transaction failures. It intelligently correlates transaction events across multiple banking systems—including the Core Banking System (CBS), Payment Gateway, NPCI Settlement, Customer Relationship Management (CRM), Audit Logs, and Incident Reports—to provide a unified view of each transaction.
Using NitroStack MCP, SPECTRA AI dynamically invokes specialized tools to retrieve relevant data, analyze transaction lifecycles, identify the exact point of failure, determine the root cause, and recommend the most appropriate operational action. Instead of manually navigating multiple dashboards and log systems, banking operations teams receive a comprehensive AI-generated investigation report within seconds.
The platform also generates operational memos, summarizes findings, and provides actionable recommendations such as transaction retries, settlement verification, refund initiation, escalation, or customer notification. By combining autonomous reasoning with secure tool-based access to enterprise data, SPECTRA AI reduces investigation time, improves operational efficiency, enhances customer experience, and minimizes financial and operational risks.
Factory downtime is not a problem with the machines. It is a problem with making decisions. When an important machine breaks down people waste a lot of time looking at manuals for the machines logs for the machines, systems for keeping track of inventory schedules for production and reports about money before they can make the right decision.
Forge-OS is a system that uses intelligence to help with big problems in the factory. It makes a copy of the factory during a crisis. This copy is called a Decision Twin. The system uses resources and tools from MCP to analyze a lot of information. This information includes logs for the machines manuals for the machines, history of maintenance, inventory how production will be affected and financial risks.
Forge-OS does not work like automated systems or tools that try to predict when maintenance will be needed. Instead it comes up with different plans to fix the problem. It then predicts how each plan will affect the factory and its money. The system shows the people in charge a list of options with reasons why they're good choices. A person must agree to the plan before it is put into action. This ensures that the people in charge are always in control and know what is happening.
Once the plan is approved Forge-OS automatically takes care of tasks. These tasks include maintenance, inventory, production, buying things and making reports. The system uses workflows that are powered by MCP. Every time there is a problem the system saves what happened into a kind of memory for the factory. This helps the factory learn and respond faster to problems in the future.
Forge-OS can help the factory go from a machine to a decision by the people in charge in less, than 60 seconds. It helps turn problems in the factory into fast and smart decisions that make sense.
Every year, thousands of students miss better educational opportunities—not because they lack merit, but because they miss critical deadlines, misunderstand complex counseling procedures, submit incorrect documents, or fail to receive timely and personalized guidance. Admission-related information is scattered across multiple official websites and notifications, making the process overwhelming for students and their families.
silicon architect is an MCP server that gives AI coding assistants a real, evidence-based semiconductor design toolchain — not simulated output. It runs actual RTL elaboration and formal bounded model checking via a WebAssembly build of Yosys, performs genuine sky130 standard-cell synthesis with measured cell area, and executes real FPGA place-and-route with nextpnr for ECP5 and iCE40, reporting achieved Fmax rather than estimates. Layered on top are IP discovery, revision history, netlist graph visualization, auto-generated design reports, and a transparent cost sheet, so an AI client can choose its next tool based on what the previous one actually produced. The result: fewer manual handoffs between specialist EDA tools, earlier detection of RTL defects before they become costly downstream, and an auditable, evidence-backed workflow that gives smaller teams and startups access to professional chip-design capability — all while keeping engineers in the loop reviewing real outputs, not AI-generated guesses.
RegPilot-Nexus is an MCP (Model Context Protocol) server that empowers AI agents to understand and reason over complex regulatory and compliance documents. Organizations struggle with fragmented regulations, manual compliance checks, and constantly evolving requirements, making compliance slow, expensive, and error-prone. RegPilot-Nexus addresses this by exposing standardized MCP tools for semantic document retrieval, obligation extraction, compliance reasoning, and real-time regulatory intelligence, enabling AI applications to automate compliance workflows with accuracy and consistency.
What it does:
LogicLens is an automated argument analysis engine that transforms complex, unstructured academic text into visual logic graphs. Using the Model Context Protocol (MCP), it enables researchers to automatically extract claims and evidence, map their relationships, and perform algorithmic audits to detect logical fallacies like circular reasoning and isolated claims.
Who it is for:
It is designed for university students, academic researchers, and peer reviewers who need to rapidly synthesize large volumes of literature and ensure the logical integrity of their arguments before publication or submission.
What makes it special:
Unlike static annotation tools, LogicLens is a protocol-native, interoperable engine. It doesn't just "highlight" text; it performs deep-graph traversal to mathematically validate reasoning chains. By exposing these insights through standard MCP tools and resources, it integrates seamlessly into any research workflow, providing a rigorous, automated "logic health check" that enhances the quality of scholarly work.
Licenix integrates with Model Context Protocol (MCP) clients like ChatGPT and Claude to streamline regulatory compliance for startup founders. By identifying and gathering required operational licenses during the ideation phase, Licenix removes legal friction before launch.
The Disaster Response Coordinator idea is compelling because it's not just one tool doing one job — it's a multi-domain situational-awareness layer for emergency management, where an LLM agent can reason across 7 normally-siloed data domains in a single conversation:
1.
Weather — cyclone/flood warnings and severity
2.
Roads — blocked routes, travel-time estimates
3.
Hospital — nearest facility, bed availability
4.
Inventory — supply levels, shortage detection
5.
Shelter — capacity, nearest-shelter lookup
6.
Population — density, vulnerable-group identification
7.
Simulation — flood-spread modeling, food-demand forecasting
Sentinel bridges the gap between passive AI chatbots and active infrastructure management. Built on the Model Context Protocol (MCP), it turns an LLM into an autonomous site reliability engineer, security analyst, and executive advisor — all in one. The system integrates real-time threat intelligence for IP reputation checks, live Grafana/Kubernetes diagnostics for outage triage, and a multi-agent "Boardroom" module where specialized CFO, CMO, and CTO agents independently assess strategic decisions using live company data. Every action that touches production — blocking an IP, rolling back a deployment — requires explicit human approval before execution, making Sentinel a demonstration of agentic AI that's both genuinely useful and safely constrained.
AI Incident Commander is an AI-powered incident response platform that helps engineering teams detect, investigate, and resolve production incidents faster. By integrating logs, deployments, commits, monitoring data, and team communication into a single intelligent dashboard, it automatically analyzes evidence, identifies potential root causes, and recommends actionable solutions. The platform reduces Mean Time to Resolution (MTTR), minimizes downtime, and enables teams to respond to critical incidents with confidence through AI-assisted investigations, real-time analytics, and a centralized knowledge base.
This Agentic AI can be used by a startup or even a multi-million dollar business as it assists the executives in crucial decision making of the company by taking into account of multiple factors such as financial viability, business risk, strategic position, industry intelligence, latest developments etc.
YatraZ is an MCP server that aids the common man in traveling and commuting from one place to another. It acts as a personal travel guide that provides bus timing and route information, pricing details and ticket booking to a commuter. Designed for the everyday commuter, YatraZ eliminates the friction of navigating fragmented transit systems. By exposing real-time route optimization, dynamic pricing, and seamless ticket booking directly to the LLM, YatraZ allows users to simply ask, 'How do I get to the central station before 9 AM for under ₹50?' and instantly receive a fully actionable, bookable itinerary.
STARK (Secure Trusted AI Response Kernel) is an MCP-powered Secure AI Space Mission Commander designed to revolutionize satellite mission operations through intelligent automation, secure decision-making, and real-time space awareness. Instead of relying on multiple disconnected systems, STARK unifies live space weather intelligence, orbital collision analysis, satellite health monitoring, mission simulation, and Zero Trust security into a single AI-driven Mission Control platform.
Built on the Model Context Protocol (MCP), STARK leverages specialized Tools to perform mission-critical analysis, Resources to access structured mission knowledge and security policies, and reusable Prompts to generate explainable mission reports and emergency response plans. Engineers can interact seamlessly through an intuitive dashboard or voice commands, while STARK analyzes mission conditions, predicts potential threats, verifies command authorization, and recommends the safest operational actions.
By combining Agentic AI, MCP, cybersecurity, and space intelligence, STARK transforms traditional satellite monitoring into an intelligent, secure, and autonomous mission command system for next-generation space operations.
Foresight exists because of a story too common in manufacturing: a supplier goes dark, a flood, a factory fire, a border shuts down, and the buyer finds out weeks later, scrambling through cold calls and outdated directories while production stalls and losses pile up. By the time a replacement is found, the damage is already done.
Foresight closes that gap. It's an MCP server that lets procurement teams find and rank alternate suppliers the moment they need one, mid-disaster or on a random Tuesday, by combining live hazard feeds (earthquakes, storms, floods, wildfires) with real compliance screening and supplier discovery from ERP data, web search, and Open Supply Hub evidence.
Built for: procurement managers and supply-chain analysts who can't afford to discover a supplier is compromised only after the damage is done.
Why it's different: instead of an alert saying "something's wrong," Foresight tells you what's wrong, who else can step in, how they compare on cost and delivery, and hands over a decision-ready memo, so the next supplier search takes minutes, not weeks.
Echo Runtime is an AI-powered Model Context Protocol (MCP) server that transforms natural language into structured GitHub workflows. Before any operation is executed, it simulates the workflow, analyzes potential risks, and generates detailed execution reports. This enables developers and teams to validate repository changes safely, reduce human error, and automate complex GitHub operations through a conversational interface.
What it does: connects directly to a startup's existing tools, analytics, CRM, finance, email, social media, and gives founders one single AI chat interface instead of ten dashboards to manage. it tracks company health like revenue, burn, runway, churn, and growth rate, watches competitors and market trends, handles founder correspondence by summarizing emails, prepping investor updates, and briefing founders before meetings, forecasts cash flow and flags financial risk, and runs growth and marketing execution like ad campaigns, social content, scheduling, and performance tracking. all of this gets synthesized into a weekly briefing with specific next action recommendations grounded in the startup's actual data, not generic advice. what makes it special is that most tools out there do one piece of this in isolation, an analytics dashboard here, an ad tool there, a scheduler, a chatbot that doesn't even know your numbers. this connects all of it into one continuous agent with real operational context, so it's not guessing, it's working off your actual data. and every action it wants to take, sending an email, publishing a post, running an ad, goes through founder approval first, so the agent proposes and the founder decides. it's not another tool you have to check, it's the one thing that tells you what to check.
FinPilot AI is an MCP-powered autonomous financial operating system designed to transform the BFSI sector through intelligent AI agents. The platform acts as a JARVIS-like financial assistant that can understand user requests, plan workflows, execute secure financial operations, and provide real-time insights.
Using a multi-agent architecture and Model Context Protocol (MCP), FinPilot AI connects banking, lending, insurance, compliance, and financial analysis workflows into a unified AI ecosystem. It automates critical processes such as KYC verification, loan eligibility assessment, fraud detection, AML compliance checks, portfolio analysis, insurance claim validation, and financial reporting.
The platform enables customers, financial institutions, and enterprises to interact with AI agents that can reason, use tools, access resources, and execute actions securely. FinPilot AI reduces operational complexity, improves decision-making, enhances fraud prevention, and delivers personalized financial intelligence.
THALETUS is an autonomous multi-agent hospital operating system designed to transform traditional healthcare management into an intelligent, connected, and self-operating ecosystem. Built using Agentic Artificial Intelligence and Model Context Protocol (MCP), THALETUS enables hospitals to move beyond conventional software systems by introducing a team of specialized AI agents that can understand objectives, reason about complex situations, collaborate with each other, access hospital resources, and execute operational workflows autonomously.
Modern hospitals are highly complex environments where multiple departments such as patient registration, doctor management, appointments, admissions, emergency care, insurance, billing, pharmacy, and diagnostics must work together seamlessly. However, most hospitals still rely on isolated software systems where each department operates independently. This creates communication gaps, increases administrative workload, delays decision-making, and negatively impacts patient experience.
A patient journey in a hospital involves multiple interconnected processes. A simple outpatient visit requires registration, doctor availability checking, appointment scheduling, medical record retrieval, consultation management, billing, and follow-up coordination. Similarly, inpatient care requires admission processing, bed allocation, doctor assignment, insurance verification, treatment tracking, and discharge management. These processes often require significant manual coordination between hospital staff, leading to long waiting times, operational inefficiencies, and increased chances of human errors.
THALETUS addresses these challenges by introducing an intelligent hospital command system where multiple AI agents collaborate as a virtual healthcare operations team. Instead of acting as a simple chatbot that only responds to queries, THALETUS uses autonomous agents capable of planning actions, selecting appropriate tools, retrieving informati
SATYA is an AI-powered institutional call verification platform that helps organizations prevent caller impersonation and protect customers from fraud. Scammers frequently exploit the names of trusted institutions such as banks, government agencies, universities, hospitals, telecom providers, and enterprises to deceive people into sharing sensitive information or making fraudulent payments. SATYA enables organizations to verify the identity of their representatives before any sensitive conversation takes place.
Authorized personnel are securely enrolled by administrators and authenticate themselves using AI-based facial recognition with head-turn liveness detection before initiating a verified call. Customers can then instantly verify the authenticity of the caller through the SATYA mobile app, which displays the representative’s verified identity, organization, designation, and active verification status in real time.
Built using NitroStack and the Model Context Protocol (MCP), SATYA exposes secure verification capabilities while integrating seamlessly with existing organizational workflows. Unlike traditional fraud detection systems that respond after an attack occurs, SATYA prevents impersonation at the start of communication by establishing trust before sensitive information is exchanged. The platform is organization-agnostic, making it suitable for any institution that wants to strengthen customer trust and reduce fraud.
Reviewing contracts—whether legal agreements, employment offers, NDAs, vendor contracts, lease agreements, or brand and influencer deals—is often slow, expensive, and inconsistent. As a result, freelancers, influencers, employees, entrepreneurs, and small business owners frequently sign legally binding documents without fully understanding the risks they contain.
Many contracts include clauses that are unfair, overly restrictive, ambiguous, or even unenforceable depending on the applicable jurisdiction. Identifying these issues typically requires legal expertise, making professional contract review costly and inaccessible for many individuals and smaller organizations.
ClauseMCP addresses this challenge by providing a universal contract intelligence engine that analyzes contracts, identifies potentially risky or unfair clauses, explains them in plain language, and delivers actionable insights—helping users make informed decisions before they sign.
DroidSec Auditor is a Model Context Protocol (MCP) server that transforms AI assistants into expert Android reverse engineers by automating penetration testing tasks like static analysis, hardcoded secret detection, and local database auditing.
Founder Copilot is a demo of what an always-on AI cofounder for Indian founders can look like — plugged into DPIIT, SIDBI, MSME, and the real programs that matter.
NyayaOS is an Agentic AI Operating System for the Justice Ecosystem that streamlines and automates end-to-end judicial workflows using specialized AI agents and Model Context Protocol (MCP) servers. Unlike traditional legal chatbots that only answer questions, NyayaOS intelligently coordinates complete legal processes, including case intake, identity verification, document processing, case registration, legal research, hearing scheduling, notifications, and AI-assisted decision support while ensuring that judges and authorized officials retain full control over all critical legal decisions. The platform is designed for citizens, lawyers, judges, court registry officials, and judicial administrators, providing each stakeholder with dedicated tools to simplify legal processes, reduce administrative workload, and improve access to justice. What makes NyayaOS unique is its modular Agentic AI architecture, where a Planner Agent analyzes user requests and orchestrates multiple domain-specific MCP servers to execute tasks efficiently, making the system scalable, explainable, secure, and easy to extend. By combining AI-driven reasoning, Retrieval-Augmented Generation (RAG), and standardized MCP-based integrations, NyayaOS transforms fragmented judicial systems into a unified intelligent platform that accelerates case processing, enhances transparency, improves legal research, and enables faster, more efficient, and accessible justice without replacing human judgment.
Hospital AI Assistant — a digital hospital system that handles patient registration, symptom-based department/doctor assignment, real-time doctor profiles, appointment booking with automated confirmations and PDF invoices, doctor-to-doctor note sharing, a centralized medical records database (history, prescriptions, labs, imaging), pharmacy inventory with prescribing, and automated report generation plus email/call notifications.
Most meeting assistants stop at notes. CloseLoop AI continues after the meeting: it reads a transcript, extracts the real decisions (not just a summary), figures out who owns each one and by when, creates the Jira ticket, schedules the calendar deadline, posts the Slack follow-up, and later chases anything that goes overdue — with zero manual handoff between "we decided X" and "it's assigned, scheduled, and being followed up on."
The Problem
Meeting transcripts contain names and natural-language deadlines ("Sarah will finish the dashboard by Friday") — never email addresses, Jira project keys, Slack channel IDs, or ISO timestamps. Every meeting-assistant tool we tried either asked the user questions a transcript can't answer, or silently guessed and got it wrong. We treated that as a schema-design problem, not a prompting problem, and fixed it at the tool-contract level.
How It Works
Meeting → Transcript → Decision Extraction → Owner Resolution → Deadline Resolution
→ Jira Ticket → Calendar Deadline → Slack Follow-up → Escalation
→ Weekly Report (Notion + Email) → Executive Dashboard
Real structured extraction via Gemini — decisions, owners, deadlines, priority, action items, risks, with a confidence score — not a vague summary.
A human-in-the-loop confidence gate: anything under 80% confidence never becomes a ticket automatically; it's routed to a Slack approval request instead.
An identity resolver that turns "Sarah Chen" into her real email/Slack/Jira/GitHub identity, backed by a small learned team-member directory, with an explicit ambiguous/not-found state instead of a silent wrong guess.
A deadline resolver that turns "Friday", "next week", "end of sprint" into an ISO date relative to when the meeting actually happened.
One-call orchestration (process_meeting) that runs the entire pipeline server-side instead of forcing the AI to manually chain 5+ tool calls.
Subscription waste hides in plain sight. Free trials silently convert to paid plans, prices creep up at renewal without anyone noticing, and forgotten subscriptions keep charging long after anyone actually uses them. Nobody manually audits their inbox to catch this — the information is scattered across dozens of emails, and checking is tedious enough that most people simply don't.
Subscription Sniper is an MCP server that solves this. A user asks one natural question — "check my subscriptions and tell me where I'm losing money" — and the agent conducts a full financial audit on its own. It connects to a real Gmail inbox via OAuth, fetches billing emails, extracts structured subscription data, and detects price increases by comparing amounts across billing cycles. It infers real usage by checking for genuine engagement signals from each sender, rather than guessing, and calculates total spend alongside potential savings. When something's flagged as waste, it generates a safe, manual, step-by-step guide to cancel UPI autopay mandates on PhonePe, Google Pay, or Paytm — the agent never auto-cancels or touches payment methods directly, keeping the user fully in control.
Built on NitroStack's MCP framework, the project uses six Tools (fetch, extract, engagement-check, calculate, cancellation playbook, and connection status), five Resources holding live structured subscription data, and two Prompts that frame the agent's reasoning — one as a financial advisor hunting waste, the other formatting alerts clearly for the user.
Live Gmail integration is fully functional and connected, demonstrated using a dedicated test account seeded with synthetic subscription emails formatted to match real-world billing patterns, including deliberate price-change and engagement-contrast scenarios.
This solves a problem nearly everyone has felt but nobody actively tracks — translating directly into real monthly savings, not just an insight report.
ClinResolve is an MCP-powered clinical decision support server that helps healthcare professionals make safer medication decisions by reconciling evidence from multiple trusted medical sources instead of relying on a single API or AI model.
Traditional clinical assistants often provide recommendations without clearly explaining how a decision was reached or how conflicting medical evidence was resolved. ClinResolve addresses this challenge through an Evidence Reconciliation Engine that collects information from multiple clinical evidence providers, evaluates agreement and conflicts between sources, incorporates patient-specific context, and generates transparent, explainable medication recommendations.
Built using the NitroStack MCP Framework, ClinResolve demonstrates the complete capabilities of the Model Context Protocol by combining Tools, Resources, and Prompts into a unified clinical workflow.
The server provides five core MCP tools:
List Patients – Browse available patient records.
Select Patient – Retrieve detailed patient profiles.
Collect Evidence – Gather medication evidence from multiple clinical sources.
Generate Decision – Produce explainable medication recommendations (Approve, Caution, or Avoid).
Audit History – Review previous clinical decisions for traceability.
MCP Resources provide structured clinical context, including patient profiles, hospital policies, clinical guidelines, drug knowledge, and historical audit records, while reusable Prompts support evidence summarization and decision explanation.
Rather than functioning as a chatbot, ClinResolve acts as an AI-powered clinical decision engine that emphasizes transparency, explainability, and evidence-based reasoning. Every recommendation is accompanied by supporting evidence, conflicting evidence, missing information, and a clear rationale, enabling clinicians to understand why a recommendation was generated before making a final decision.
By leveraging the Model Context Protocol, ClinRe
The MCP Server aims to connect the backend which uses numerous ai agents such as cursor and claude to analyze a transaction from a user. Currently, it works on a claude based front end with each interaction made manually. It can be improved upon by making a tool to directly feed bank transactions into the prompting cursor, so that the workflow is automated. The user may request to create the user and store transactions, along with other transaction information such that fraudulent activity can be detected based off previous transactions. We can also choose to see the list of users in our claude front-end's memory, and if the user wishes to be wiped off the memory, the user profile can be deleted by telling it to do so. The MCP takes in the transaction as text input and then is given a transaction and risk profile, along with a score and feedback telling the user what the next recommended action is.As such the user can choose whether to clear the transaction, block it or just keep it for review later. We thus seek to reduce bank fraud using our ai driven mcp.
Nexus is a premium multi-agent education and research platform built on NitroStack, designed to showcase how intelligent agents can collaborate to transform academic workflows. It features a polished command-center UI and five specialized agents—Research, Learning, Debate, Review, and Synthesis—working together to explore sources, build study paths, frame opposing viewpoints, review drafts, and create action-ready summaries. With additional tools for concept mapping, quiz outlines, research summaries, and execution checklists, Nexus is positioned as an innovative hackathon demo that makes learning, teaching, and scholarly discovery feel structured, professional, and future-ready.
Shift-Left FinOps is an autonomous infrastructure planning agent that designs cloud infrastructure by reasoning about cost, performance, scalability, reliability, and organizational policies before generating Terraform.
Financial institutions employ armies of analysts to manually investigate flagged transactions, which is slow, expensive, and prone to human error. An autonomous MCP server acts as a forensic accountant; When a transaction is flagged, the agent dynamically queries external databases, maps corporate ownership structures, and compiles a comprehensive compliance dossier.
Natural disasters often create a flood of information at the worst possible time. Emergency responders have to monitor weather updates, identify affected areas, locate hospitals and shelters, coordinate rescue teams, track available supplies, and make critical decisions—all while every minute counts. Most of this information is scattered across different platforms, forcing responders to switch between multiple tools and manually piece together the situation. We wanted to simplify that process.
DRIP (Disaster Response Intelligence Platform) is our attempt at bringing disaster response into a single intelligent ecosystem. Instead of treating every emergency task as a separate problem, DRIP breaks the response into specialized AI modules that work together under one central planner. Each module focuses on a specific responsibility—gathering live disaster intelligence, locating hospitals and shelters, managing rescue vehicles and volunteers, tracking relief resources, or planning operations. The Planner coordinates these modules and combines their outputs into a clear, actionable Situation Report that emergency teams can use immediately.
Our goal wasn't just to automate tasks, but to improve decision-making under pressure. By integrating real-time information with AI-assisted planning, responders spend less time searching for information and more time acting on it. The modular MCP architecture also makes the platform flexible, allowing new emergency services or data sources to be added without redesigning the entire system.
We believe DRIP has the potential to support disaster management authorities, emergency operation centers, NGOs, and first responders by reducing coordination delays, improving resource utilization, and providing better situational awareness during critical operations. In high-pressure emergencies, even saving a few minutes can translate into saved lives, and that's the impact we set out to create.
BruteForce is an AI-orchestrated compliance platform that unmasks Ultimate Beneficial Owners(UBO) hidden behind layered offshore shell companies. It enforces a strict architectural boundary: a deterministic graph core handles all ownership math, entity resolution, and sanctions screening with zero LLM involvement, while an autonomous AI planner coordinates the investigation - selecting tools, evaluating evidence sufficiency, and compiling audit-ready dossiers. If it didn't come from a deterministic tool, it's not a fact.
Adaptive Infrastructure Resilience MCP (AIR-MCP) is a production-inspired, autonomous decision orchestration platform designed to manage and mitigate critical incidents in smart datacenters and grids.
Using a multi-agent orchestration engine, it actively monitors telemetry, assesses operational risks, migrates workloads, schedules technicians, and procures replacement parts to restore cooling loops and protect service level agreements (SLAs).
GreenOps One is an AI-powered Autonomous Sustainability Operating System designed to optimize the environmental and operational efficiency of modern AI data centers. Built using the NitroStack MCP framework, it intelligently orchestrates multiple sustainability intelligence modules to analyze live environmental conditions, energy usage, carbon emissions, water availability, renewable energy potential, battery utilization, and AI workload distribution.
The system integrates real-time data from trusted public sources such as weather, solar, climate, air quality, and energy APIs to generate intelligent optimization decisions. By combining these live insights through an MCP orchestration pipeline, GreenOps One recommends actions such as enabling free-air cooling, shifting GPU workloads to off-peak hours, prioritizing renewable energy, optimizing battery charging, reducing water consumption, and improving ESG performance.
The platform features an interactive Mission Control dashboard that visualizes the complete optimization process through a live orchestration pipeline, AI decision engine, sustainability reports, impact cards, before-and-after comparisons, and optimization results. This enables operators to monitor infrastructure health, understand sustainability impacts, and make informed operational decisions in real time.
GreenOps One demonstrates how AI agents and MCP architecture can work together to create greener, more energy-efficient, cost-effective, and environmentally responsible AI infrastructure. By transforming fragmented operational data into actionable intelligence, the platform helps reduce carbon emissions, improve renewable energy utilization, optimize cooling strategies, and enhance overall sustainability while maintaining high-performance AI operations.
AutoGuard is a full-stack, AI-driven predictive maintenance and operations dashboard designed for industrial manufacturing plants. It monitors machine health in real-time through telemetry data, uses a FastAPI-hosted XGBoost machine learning model to predict failure probabilities and classify risk levels, and employs a TypeScript-based Model Context Protocol (MCP) server to coordinate incident responses. When a high-risk failure is detected, the system either auto-dispatches an available technician with a work order (if risk is 100%) or creates a manager-facing pending approval widget (if risk is elevated but below 100%), facilitating automated spare parts inventory checks, downtime cost analysis, and a seamless manual approval workflow to minimize manufacturing downtime.
The idea is to create an AI-powered platform that transforms traditional 2D house plans into a fully interactive 3D digital home. Users simply upload their architectural plans, and the AI automatically generates a realistic 3D model of the house
Construction is India's second-most-hazardous sector, averaging an estimated 38 fatal accidents a day — and that likely undercounts the real toll, since ~30% of the workforce is unregistered. 60% of these deaths are falls from height and 25% are structural collapses — hazards a camera can catch, if anyone's watching. But the workforce is largely migrant, multilingual, and low-literacy, so an English dashboard alert never reaches the person in danger.
SurakshaMCP is the first voice-first, multilingual Construction-Safety MCP server. Feed it a site photo or CCTV frame, and it detects PPE/hazard violations with a fine-tuned YOLOv8 model, identifies the worker's spoken language from audio, and delivers a spoken safety alert in their mother tongue — in one of 10 Indian languages via AI4Bharat Indic-TTS. This isn't a translated dashboard; it's an actual voice warning, load-bearing for real-time hazards like a worker straying near moving machinery.
Every detection persists to a SQLite incident DB and rolls up into a BRSR (SEBI ESG) Principle-3 safety report — LTIFR, training coverage, monthly trends — so India's MSME contractors can hand large builders the safety documentation they increasingly need to win bids. Alerts can also reach a foreman's phone via WhatsApp (Twilio).
Architecturally, SurakshaMCP is MCP-native and host-agnostic: safety intelligence is exposed as Tools, Resources, Prompts, and a long-running Task (run_site_safety_audit), built on NitroStack (TypeScript), wrapping a Python FastAPI inference service (YOLOv8 + language-ID + TTS). Any MCP host — Claude, ChatGPT, Copilot Studio, even a WhatsApp bot — can orchestrate it. Enterprise camera-safety systems (viAct, Intenseye) are closed, English-first dashboards priced for large firms. SurakshaMCP is MSME-affordable — no edge hardware, just a phone photo — and composable into whatever workflow a small contractor already uses.
OmniMCP constitutes a universal hardware capability layer that allows AI agents to interface with industrial devices via the Model Context Protocol (MCP). Rather than requiring developers to construct bespoke integrations for each individual machine, the framework permits hardware to be connected through lightweight adapters that expose standardised capabilities, among them machine control, image inspection, and sensor data acquisition. AI applications are thus able to make use of these capabilities without any need to understand the underlying hardware-specific protocols, which in turn facilitates rapid deployment across smart factories, industrial automation systems, and Industry 4.0 environments more broadly.
The viability of this approach is demonstrated by means of a prototype comprising an ESP32 machine controller together with an inspection camera. The underlying architecture, moreover, has been designed with extensibility in mind: it is intended to accommodate PLCs, robots, CNC machines, sensors, and other industrial equipment through what remains, from the perspective of the AI application, a single unified interface.
Researchers spend a significant amount of time searching for papers, comparing methodologies, identifying research gaps, organizing references, and planning experiments. Existing AI tools primarily assist with summarization, leaving much of the research process manual, fragmented, and time-consuming. And our solution introduces an agentic approach to research, where multiple AI agents collaborate to automate and optimize every stage of the research lifecycle. The platform not only retrieves and summarizes information but also reasons over existing literature, uncovers unexplored opportunities, recommends research directions, and generates actionable research plans.
Aegis AI – Student Briefing is an AI-powered college assistant that provides students with instant access to academic information, schedules, notices, and personalized assistance through natural language.
An MCP agent that catches trade-based money laundering (over/under-invoicing, phantom shipments) in India's export-import finance flow. It cross-references declared invoice values against independent commodity-price benchmarks, checks FEMA 2026 realization-timeline compliance via EDPMS, and flags counterparties reusing suspiciously identical invoice values across transactions. When red flags accumulate, an LLM reasoning layer drafts a ready-to-file Suspicious Transaction Report — not just a raw alert. Built for AD-bank compliance desks, right as RBI's new FEMA rules (Oct 2026) impose penalties up to 300% of transaction value. Uses all three MCP primitives: Resources, Tools, Prompts.
PortfolioPulse is a portfolio-aware MCP server that helps investors and advisors transform scattered investment holdings into a clear, actionable portfolio view. Investors commonly hold assets across broker accounts, mutual funds, crypto platforms, bank statements, and digital-gold providers. While these platforms show individual positions, they rarely explain total allocation, concentration risk, portfolio-specific news, or the policy developments that may matter to the assets a person actually owns. Generic market news creates more noise because it is not connected to the user’s holdings.
PortfolioPulse solves this through a set of MCP tools built with NitroStack. It loads and normalizes portfolio holdings, calculates total value and unrealized profit or loss, summarizes asset-class and sector allocation, identifies top holdings, and highlights concentration risks using transparent rule-based checks. For example, its aggressive sample portfolio clearly surfaces a high crypto allocation and a concentrated ETH position, helping users understand the source of portfolio volatility rather than simply seeing a return number.
The server also delivers relevant news filtered by the investor’s held symbols and sectors. Each item includes a plain-language explanation of why it matters to that portfolio. Additional tools support educational rebalancing suggestions and regulatory alerts relevant to the investor’s exposures.
Built as an MCP server, PortfolioPulse can be used by compatible AI clients and embedded in advisor, dashboard, or personal-finance workflows. Its potential impact is to make portfolio reviews faster, more transparent, and more relevant: investors can better understand exposures, while advisors can reduce time spent manually combining information across accounts. PortfolioPulse is for informational and educational use only and does not provide investment advice.
MCP Execution Ledger is a durable execution layer for MCP-based AI agents that improves reliability, debugging, and operational transparency. It intercepts every tool invocation and records it in a single immutable execution log, creating checkpoint markers at completed tool boundaries. From this unified execution history, developers can safely resume interrupted workflows, inspect execution timelines through read-only replay, and generate audit reports without modifying existing tools. The MVP also demonstrates a minimal idempotency guard for controlled workflows, preventing duplicate side effects during crash recovery. MCP Execution Ledger provides reusable infrastructure that helps MCP-based AI systems become more reliable, observable, and maintainable.
Tap & Save helps users achieve their savings goals by automatically separating their money into daily spending, savings, and emergency wallets. Users can set a savings target and a lock date to prevent impulsive spending.
The platform offers Normal Mode, where savings can be withdrawn using a PIN, and Strict Mode, where the savings wallet remains locked until the goal date. In emergencies, users can upload proof (such as a hospital bill). The AI analyzes the document, recommends whether the request is valid, and after user confirmation, simulates a direct payment to the verified merchant.
Powered by NitroStack MCP, the platform can automate spending alerts, analyze uploaded documents, manage goal deadlines, and provide intelligent financial guidance. Unlike traditional budgeting apps, Tap & Save actively helps users stay committed to their savings goals.
SmartCampusAI is an AI-powered campus assistant built using React and MCP (Model Context Protocol) on NitroStack. It helps students and faculty access campus services such as attendance tracking, classroom availability, faculty search, library book search, assignment reminders, complaint management, campus navigation, and real-time campus updates through an intelligent conversational interface.
ICU Guardian AI is a healthcare monitoring web application for ICU patient supervision. It provides a modern command-center dashboard with patient cards, detailed vitals, ECG visualization, nurse action checklists, doctor review pages, patient history timelines, and system-status placeholders.
ProofYield is an MCP-native autonomous DeFi treasury agent that researches yield opportunities across protocols, evaluates risk and policy constraints, explains every decision, executes verified on-chain transactions, and generates an auditable Decision Receipt. It brings transparency, governance, and trust to AI-powered financial automation.
CityAI – Agentic AI-Powered Smart Road Infrastructure Assistant
Overview
CityAI is an Agentic AI-powered smart city solution designed to improve the way citizens and municipal authorities handle road infrastructure issues. Every day, people encounter potholes, road cracks, and damaged road surfaces that can lead to accidents, vehicle damage, and traffic disruptions. Unfortunately, existing reporting systems are often slow, fragmented, and lack transparency, making it difficult for citizens to know whether their complaints are being addressed.
CityAI bridges this gap by combining Agentic AI, Computer Vision, MCP (Model Context Protocol), and route optimization into a unified platform that enables intelligent road damage reporting, complaint tracking, safer navigation, and data-driven infrastructure management.
Problem Statement
Road infrastructure issues remain one of the most common urban challenges. Citizens often struggle to report road damage efficiently, while municipal authorities receive large numbers of complaints without a clear mechanism for prioritization.
Current systems face several limitations:
Manual complaint registration processes
Lack of transparency regarding complaint status
No automated assessment of damage severity
Inefficient prioritization of repairs
Limited visibility into city-wide infrastructure health
No route recommendations that consider road conditions
As a result, dangerous road conditions remain unresolved for longer periods, creating risks for commuters and reducing public confidence in civic reporting systems.
Our Solution
CityAI introduces an intelligent citizen-to-gove
ahayakAI is a voice-first AI banking assistant that helps people complete bank forms through simple conversations. It is designed especially for elderly, low-literacy, and first-time banking users who struggle with reading, writing, or navigating complex banking forms. Instead of manually filling forms, users simply speak their details, and SahayakAI extracts the information, confirms it through voice, and generates a print-ready bank form for them to sign and submit. This makes banking more accessible, independent, and less intimidating while reducing errors. Although our current MVP supports English . Our future vision is to support multiple Indian languages and bank forms across different banks.
Aptum helps college students choose the right laptop for their studies. Simply tell it your college and branch/department, and it will analyze your curriculum to understand the software, programming languages, and tools you'll use throughout your course. Based on those requirements, it recommends the ideal laptop specifications—including processor, RAM, storage, graphics, display, and battery life—and suggests laptop models across different budget ranges to ensure you're well-equipped for your entire degree.
MCP-first agentic penetration-testing system. Two MCP servers built on NitroStack expose every capability as typed, scope-checked tools. Four thin agents sequence them; NitroStack is the control plane.
Pulse gives AI the context it’s been missing: you. Through a Pixel Watch, companion app, and open NitroStack MCP server, Pulse connects your speech and heart rate to AI agents in real time. It can recover your next line, detect moments of stress, summarize conversations, and reveal how your confidence changes throughout a presentation. Unlike closed smartwatch platforms, Pulse keeps your data encrypted, user-controlled, and open to any AI experience developers can imagine.
This is a Farmer Advisory App designed to assist agricultural professionals with:
Available Features:
Weather-Based Crop Recommendations - Get crop suggestions based on your location's weather and market prices
Fertilizer Prices & Advice - Get current market fertilizer prices and recommendations for specific crops in your district
Irrigation & Fertilizer Guidance - Personalized advice based on weather conditions and your chosen crop
Trusted Seed Sources - Find certified seed dealers and KVK centers for your crop and season
Farmer Schemes - Discover government schemes and benefits available in your state
Temperature Conversions - Convert between Celsius and Fahrenheit
Basic Calculations - Perform arithmetic operations
NutriGuide AI is an MCP-powered intelligent nutrition assistant built with Nitro Studio. It analyzes user health information and blood biomarkers to generate personalized nutrition reports, meal plans, grocery lists, and lifestyle recommendations using an agentic AI workflow.
AIRA is an AI-powered manufacturing intelligence platform that helps industries quickly identify the root causes of production failures, quality defects, machine breakdowns, and process deviations. Instead of relying on manual investigations that are often slow and inconsistent, AIRA uses Artificial Intelligence, Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), and a manufacturing knowledge base to analyze production data, machine logs, sensor readings, maintenance records, and operator observations.
The system generates a ranked list of possible root causes, explains the reasoning behind each prediction, and recommends industry-standard corrective and preventive actions (CAPA). It also estimates the impact of each issue on production, quality, cost, and downtime, enabling engineers to make faster and more informed decisions.
By combining AI-driven analysis with explainable recommendations, AIRA reduces troubleshooting time, minimizes production losses, improves product quality, and supports continuous improvement in smart manufacturing environments. The platform is designed to integrate seamlessly with Industry 4.0 systems and provides an intuitive interface for engineers, supervisors, and plant managers to investigate failures and implement effective solutions.
Simulation MCP is a Model Context Protocol (MCP) server that bridges Large Language Models (LLMs) such as ChatGPT, Claude, and Gemini with industry-standard engineering software, enabling users to perform CAD design, molecular modeling, and engineering simulations through natural language.
Instead of requiring expertise in multiple specialized software packages, users simply describe their engineering problem in plain English. Simulation MCP intelligently interprets the request, generates an execution plan, selects the appropriate software, automates the workflow, executes the required analysis, and returns interactive visualizations, numerical results, and AI-powered explanations.
The platform is designed around a modular MCP architecture where each engineering application operates as an independent service. This enables seamless integration of multiple engineering domains while eliminating the need to manually switch between software, configure solvers, or prepare simulation models.
By combining AI planning, automated software orchestration, and professional engineering tools under a single MCP interface, Simulation MCP significantly reduces the complexity of engineering design workflows, allowing students, researchers, and engineers to focus on innovation rather than software operation.
Talos is an AI-powered cybersecurity assistant and brute-force detection platform built for defensive security analysis, incident response, and practical security education. It combines machine-learning based authentication log analysis with a broad security toolbox for passive website scanning, CVE lookup, IP investigation, password auditing, JWT decoding, phishing/link-safety checks, academic research discovery, local resource search, report generation, alerting, and mitigation planning.
What makes Talos special is its NitroStack-powered MCP server, which exposes these security capabilities as structured Model Context Protocol tools. Instead of giving generic text-only answers, AI assistants can call the correct Talos tool directly, stream live tool activity, generate evidence-backed reports, and move from detection to investigation and remediation in one workflow.
The system uses a Python security backend with a TypeScript/NitroStack MCP layer, making it suitable for local development, cloud hosting, and AI-integrated security operations. Talos has strong market potential as organizations increasingly need AI-native security tools that make threat detection, investigation, reporting, and remediation faster, more accessible, and easier to operationalize.
FinAnalyst AI is an AI-powered financial analysis platform built using the Model Context Protocol (MCP) to make stock analysis faster, simpler, and more accessible.
The platform connects to live financial market data and provides investors with key company and stock insights through specialized MCP tools. Users can enter a stock symbol such as AAPL and instantly retrieve company information, current stock price, sector and industry details, financial performance metrics, and technical indicators.
Our system analyzes important factors including quarterly revenue growth, EBIT margins, 50-day moving averages, and 200-day moving averages. These insights can be presented through an interactive Financial Analyst dashboard to help users understand a company's financial health and market momentum.
The project is designed for retail investors, students, and anyone who wants to understand stocks without manually analyzing large amounts of financial data.
What makes FinAnalyst AI unique is its MCP-based modular architecture. Instead of being a simple static finance dashboard, it exposes specialized financial tools that AI agents can discover and use dynamically. The system integrates live Yahoo Finance data with MCP tools and an interactive widget-based interface, creating a foundation for AI-powered investment analysis.
The goal is to make financial analysis more understandable, efficient, and accessible while demonstrating how AI agents and MCP can be applied to real-world FinTech problems.
COS-MCP is an MCP (Model Context Protocol) server that provides AI-powered organizational continuity planning. It maintains a knowledge graph of employees, systems, projects, and relationships, and exposes tools, resources, and prompts for knowledge graph visualization, risk analysis, employee transition planning, and organizational knowledge base queries
MediSync AI is an AI-powered Emergency Health Passport that securely stores critical patient information such as blood group, allergies, medications, and emergency contacts. During emergencies, it uses AI and MCP to instantly retrieve patient records, analyze the situation, and provide immediate first-response guidance, enabling faster and more informed medical decisions.
FutuTrust is an AI-powered loan assistant that makes applying for a loan as simple as having a chat.
Instead of filling out confusing paperwork, you just state what you need in plain English (like "I need a mortgage for a new house" or "I need to pay for college").
Behind the scenes, a Main Router Agent listens to your request and instantly hands you off to one of four specialized expert mini-agents: Personal, Home, Vehicle, or Educational Loans. These mini-agents automatically crunch the numbers, check your criteria, and show you an instant approval status on a clean visual dashboard.
Mārga is an AI-driven workflow intelligence platform that helps entrepreneurs navigate business registration, regulatory compliance, and government approvals through a collaborative multi-agent AI system. It automates workflow planning, identifies required licenses and documents, optimizes approval sequences, and provides actionable recommendations for faster business setup.
The Problem
DFIR teams rely on a fragmented toolbox: Volatility for memory forensics, tshark for network capture, exiftool for metadata, objdump for binary analysis, binwalk for carving, steghide for steganography. No two machines carry the same set — jump-boxes, SOC laptops, cloud sandboxes, and honeypot collectors all differ. Existing fixes either hardcode brittle integrations that crash when a tool is missing, or force bloated all-in-one distros just for consistency. Connecting a private, self-hosted LLM to this evidence via MCP today means hand-wiring the integration per machine — if a utility is absent, the server breaks, causing costly delays during time-sensitive incidents.
The Solution
This framework adds a runtime capability-discovery layer that flips the integration model. At startup, the server inspects the host — probing binaries, environment paths, and container engines, including health checks like verifying an operational Docker daemon instead of a fragile host Python install — to learn exactly what's usable.
It then runs three automated steps: **Discovery & Verification** (scan and validate local or containerized tools), **Dynamic Mapping** (map each tool's inputs/outputs to standard MCP definitions), and **Context Assembly** (self-assemble matching Tools, Resources like magic-byte signatures and case logs, and triage Prompts).
Missing utilities are skipped gracefully, never crashing the server. Evidence never leaves the local perimeter, making it suited to air-gapped, high-compliance environments. Because it's built on NitroStack's decorator-based SDK, this whole pattern can itself become a reusable template — scaling to new tools or commercial forensic suites needs only a manifest update, not a rebuild.
CampusMind AI is an AI-powered academic copilot that helps students learn smarter instead of studying harder. Unlike traditional AI chatbots that forget every conversation, CampusMind AI uses the Model Context Protocol (MCP) to maintain a persistent academic memory for each student. It remembers what a student has studied, identifies weak concepts, tracks progress over time, and proactively recommends what should be reviewed before exams.
The platform combines a modern web application with a NitroStack MCP server. Students can interact through an AI chat interface, view their study dashboard, access a memory timeline, manage revision plans, upload notes, monitor learning analytics, and use Exam Mode to generate personalized revision strategies based on their academic history.
CampusMind AI automatically performs actions such as:
Remembering previously studied topics
Recalling old doubts using natural language
Identifying weak concepts
Generating personalized study plans
Scheduling spaced repetition reviews
Tracking learning progress and study streaks
Creating AI-assisted exam preparation plans
What makes CampusMind AI unique is that it acts as a long-term academic companion rather than a one-time chatbot. Every interaction improves the student's academic profile, enabling the AI to provide increasingly personalized guidance over time. By combining persistent memory, intelligent planning, and proactive recommendations, CampusMind AI helps students retain knowledge, reduce exam stress, and improve learning outcomes.
Built with NitroStack MCP, Next.js, TypeScript, SQLite, and AI-powered academic memory, CampusMind AI demonstrates how persistent AI agents can transform personalized education.
Problem:
Fraud and security analysts spend most of their shift on repetitive, manual investigation — pulling device/geo data, correlating account activity, and cross-referencing logs across disconnected tools before they can even make a call. When AI is introduced to speed this up, it usually acts as a black box: a risk score with no visible reasoning. In regulated industries like BFSI and security operations, that's a real liability — decisions need to be explainable and defensible to auditors, regulators, and customers, not just "the model said 87% risk."
Solution:
Lucid is an MCP-based AI investigation agent, built end-to-end on NitroStack, that runs a full triage pipeline on any flagged alert or transaction:
Enrich — pulls device risk, geo-anomaly, and reputation signals for the alert
Correlate — surfaces related account activity (logins, resets, prior transactions)
Classify — returns a verdict (clear / flag / escalate) with a confidence score and a step-by-step reasoning chain
Escalate — auto-drafts a structured handoff packet for human (L2/analyst) review
Unlike typical AI copilots, every verdict Lucid produces comes with a fully traceable evidence chain — orchestrated as live, inspectable tasks rather than a single opaque model call — so an analyst or auditor can see exactly which data point led to which decision, not just trust a number.
Impact:
What normally takes an analyst 10–15 minutes of manual cross-tool investigation happens in seconds, with a defensible, replayable reasoning trail attached, built for the level of accountability that regulated, high-stakes decisions actually require.
Trust.dink is an MCP server that acts as a real-time compliance firewall for autonomous AI payment agents. As protocols like Google AP2 and Mastercard Agent Pay let AI assistants sign and execute transactions without human approval, they open up new risks — tampered carts, compromised agent keys, and unchecked spend. Trust.dink plugs directly into an LLM's tool-calling layer (built and demoed on Claude Desktop) to catch these risks before a transaction ever reaches a payment network.
It's built for banks, fintechs, and payment platforms that need to let AI agents transact on their customers' behalf while still meeting regulatory expectations — including India's RBI draft Model Risk Management Guidelines 2026, which mandate explainability, auditability, and real-time kill switches for autonomous financial systems.
What makes it different from a typical fraud-score API:
Cryptographic mandate verification — Ed25519 signature checks on every transaction payload, so a tampered cart (e.g. an amount silently boosted from ₹250 to ₹25,000) fails verification instantly instead of relying on heuristics alone.
Explainable risk scoring — a transparent, rule-based engine scoring signature validity, merchant scope, velocity, and geo-anomalies, rather than a black-box number.
Kill switch with immutable audit trail — a compliance officer can suspend a compromised agent's signing key instantly, blocking all further transactions, with every event chained via SHA-256 hashing for a tamper-evident log.
Live market-aware risk — pulls real-time NSE and FX data so risk thresholds tighten automatically during sector volatility or market shocks, instead of using static limits.
Trust Graph visualization — renders the full mandate chain live as a widget, so a human can see why a transaction was trusted or blocked, not just the verdict.
Trust.dink turns compliance from a bolt-on dashboard into a protocol-level safeguard AI agents literally cannot transact around.
CredEdge is an MCP server built for the BFSI & FinTech track that gives an AI agent real, live access to the core functions a bank employee performs every day — not a chatbot that talks about finance, but one that actually acts on real account data.
It exposes six tools: real-time fraud risk scoring (analyzes a transaction against the account's actual spend history — amount, location, velocity, merchant pattern — and returns a fraud probability with a concrete next-action plan), CIBIL score lookup, personal loan eligibility (explains exactly why a loan is approved or rejected and lists the documents needed — aligned with RBI's Digital Lending Guidelines on rejection transparency), SME business loan assessment (turnover, GST compliance, years in operation), insurance claim investigation (flags early claims, high-value claims, and repeat-claimant patterns), and personalized financial advice (turns income/debt data into a concrete monthly savings and investment plan). It also pulls live market data via Alpha Vantage as a real external data signal.
Every tool result renders as a polished visual widget inside the AI chat itself — risk gauges, eligibility cards, CIBIL score dials — with one-click PDF/Word export for compliance record-keeping.
Who it's for: banks, NBFCs, and fintechs that want to give an AI assistant safe, auditable access to lending and fraud decisions without exposing raw databases — the AI only ever sees what a tool explicitly returns.
What makes it special: because it's built on MCP rather than a one-off API, the exact same server works unmodified inside Claude, ChatGPT, or any future MCP-compatible banking assistant — one build, many front doors. Every decision is explainable, since a resource exposes the underlying rulebook, so an examiner can ask why a decision was made and get the exact rule that fired — not a black box.
Solar Sathi is an AI-powered Model Context Protocol (MCP) application that helps homeowners evaluate rooftop solar adoption. Using natural language, users can estimate solar generation, installation costs, government subsidies, return on investment (ROI), environmental impact, and receive personalized recommendations. The system combines live solar data, location intelligence, financial calculations, and AI-powered reasoning into a single conversational assistant.
AfterLife AI is an AI-powered estate administration copilot that helps families navigate the complex legal, financial, and administrative processes that follow the loss of a loved one.
Users simply describe their situation, and the platform analyzes the event, identifies required services, generates a personalized workflow, detects missing information, recommends asset-discovery actions, and prepares relevant documents.
The system guides users through critical tasks such as obtaining legal heir certificates, bank account transfers, pension closure, insurance claims, and property mutation. By combining AI-powered reasoning, workflow orchestration, document generation, and case tracking into a single dashboard, AfterLife AI transforms a confusing and stressful process into a structured, step-by-step experience.
Key Features
• Intelligent Death Event Analysis
• Personalized Workflow Generation
• Missing Information Detection
• Asset Discovery Guidance
• Legal & Administrative Document Generation
• Case Progress Tracking Dashboard
• AI-Powered Decision Support
Target Users
Families, legal advisors, estate administrators, financial institutions, and government service facilitators involved in post-death administrative processes.
Impact
AfterLife AI reduces confusion, saves time, minimizes procedural errors, and helps families complete critical administrative tasks more efficiently during one of the most difficult periods of their lives.
PharmaGuard AI Pro is an AI-powered clinical intelligence and pharmacovigilance platform that analyzes medical prescriptions to detect dangerous drug interactions, assess adverse drug event risks, and provide evidence-based clinical recommendations. Designed for healthcare professionals, pharmacists, hospitals, researchers, and pharmacovigilance teams, it combines prescription analysis with pharmacovigilance metrics such as PRR, ROR, and BCPNN to deliver actionable medical insights. The platform generates professional clinical intelligence reports, visualizes drug relationships through interactive knowledge graphs, and offers AI-driven risk scoring with emergency alerts to improve medication safety. What makes PharmaGuard AI Pro unique is its integration of AI-powered clinical reasoning, drug safety analytics, research intelligence, and MCP-based architecture into a single production-ready solution, enabling smarter and safer healthcare decisions in real time.
The Problem: The Edge is a Black Box
Managing remote edge devices—IoT gateways and digital kiosks—is a logistical nightmare. When nodes drop offline, engineers must manually SSH in and grep through cryptic kernel logs to guess the failure. This archaic process is unscalable, and feeding raw OS errors to standard AI agents often causes hallucinations, rendering them useless for low-level infrastructure management.
The Solution: EdgeSentinel
EdgeSentinel is a secure, autonomous, bare-metal Linux Site Reliability Engineering (SRE) Copilot. It fundamentally shifts edge diagnostics from reactive "guess and check" to intelligent, verifiable incident response, powered entirely by local AI. EdgeSentinel in Action
Unlike naive scripts that crash upon hitting missing dependencies (like nmcli), EdgeSentinel is a capability-aware platform. It dynamically maps host capabilities at startup, seamlessly falling back to portable alternatives like ip link or reading /sys/class/net directly from the kernel if needed. It never dumps raw bash errors; instead, it intercepts failures and normalizes data through strict Zod-validated contracts. Operators instantly see deterministic states (e.g., "Degraded") on a custom Cyberpunk-styled React telemetry HUD, complete with exact hardware observations and safe remediation steps. Enterprise-Grade Security & Zero-Trust
Powered by the Model Context Protocol (MCP) and local LLMs via Ollama, EdgeSentinel ensures zero sensitive system telemetry ever leaves the physical device. It operates on a strict human-in-the-loop safety model: it is read-only by default, and mutating actions (like restarting a service) require explicit human approval after a read-only preflight check. Conclusion:
EdgeSentinel is an "SRE in a box". It neutralizes command injection risks, eliminates hallucination-inducing OS errors, and bridges the gap between low-level kernel diagnostics and agentic AI to build unbreakable edge infrastructure.
FactoryLens – Because Every Second of Downtime Costs More Than Just Time
FactoryLens is an AI-powered smart factory maintenance assistant built using the Model Context Protocol (MCP) and the NitroStack SDK. It brings together machine telemetry, maintenance history, equipment manuals, and production schedules into a single intelligent interface, allowing AI assistants to access factory data through structured tools and resources.
Instead of manually investigating equipment failures, technicians can simply ask questions like "Why is Machine MCH-004 overheating?" or "What does error E101 mean?" FactoryLens automatically analyzes sensor data, reviews maintenance records, consults equipment manuals, identifies the most likely root cause, estimates business impact, and can even generate a maintenance work order for critical issues.
Rather than displaying raw data, FactoryLens connects information from multiple sources to provide clear, evidence-backed insights through interactive dashboards and investigation reports. This helps technicians, engineers, and plant managers diagnose problems faster and make informed maintenance decisions.
By reducing troubleshooting time, minimizing equipment downtime, and streamlining maintenance workflows, FactoryLens improves factory productivity and reliability. As it evolves with live IoT integration, predictive maintenance, and cloud connectivity, it has the potential to enable smarter, more proactive industrial operations.
FactoryMind AI is an AI-powered smart manufacturing assistant built using NitroStack and the Model Context Protocol (MCP). It enables users to interact with factory systems using natural language to monitor machine health, analyze equipment, predict maintenance needs, check spare part availability, assign technicians, create maintenance tickets, and visualize production KPIs. By combining NitroStack's AI orchestration with MCP tools, FactoryMind AI streamlines industrial operations, automates workflows, and enhances decision-making in Industry 4.0 environments.
CodeWeaver is an MCP server that turns free-text clinical notes into ranked ICD-10-CM diagnosis code suggestions, with transparent, auditable reasoning behind every result.
What it does: Given a clinical note, CodeWeaver segments it into sentences, detects negated or ruled-out conditions so they're never miscoded, and fuzzy-matches the rest against real CMS FY2026 ICD-10-CM data (1,934 codes across cardiac, diabetes, respiratory, and a curated musculoskeletal slice). Candidates then run through a constraint engine enforcing official Excludes1 rules (hard-blocking mutually exclusive diagnoses, like Type 1 and Type 2 diabetes charted together), Excludes2 rules (flagging codes that can co-occur), specificity ranking, and 7th-character extension requirements. When nothing scores confidently, CodeWeaver says so explicitly and live-queries the NLM Clinical Tables API as a clearly-labeled, unvalidated fallback rather than guessing.
Who it's for: Medical coders and clinical documentation teams who need a fast first pass on note-to-code mapping, with every suggestion traceable to the exact phrase and CMS rule behind it — built for coder review, not autonomous decisions.
What makes it special: Most coding-assist demos wrap an LLM with no real constraint logic. CodeWeaver uses actual CMS source data, parsed directly from the official FY2026 XML tabular index, and implements the rules that make ICD-10 genuinely hard — Excludes1/2 semantics, specificity penalties, 7th-character disambiguation — as explicit, explainable logic. It names its exact scope, refuses to present low-confidence matches as suggestions, and clearly separates validated local results from unvalidated external lookups.
CodeWeaver is a coding-assistance tool, not a diagnostic one — output requires review by a qualified coder. Clinical notes used for testing are synthetic; the ICD-10-CM data is real.
HelmsMan is an autonomous Kubernetes remediation platform built using the Model Context Protocol (MCP) and the NitroStack MCP SDK. It enables AI agents to safely analyze, reason about, and optimize a live Kubernetes cluster while ensuring that every infrastructure change complies with Kubernetes availability policies. Traditional autoscaling mechanisms rely mainly on CPU or memory thresholds and cannot determine whether a scaling operation is operationally safe, often leading to over-provisioned clusters, increased cloud costs, or actions that risk application downtime. HelmsMan addresses this limitation through a multi-agent architecture consisting of a FinOps Agent, which identifies cost-saving opportunities by detecting idle or over-provisioned workloads, and an Availability Guardian, which evaluates the potential impact of every proposed action on application reliability. Instead of allowing AI agents to directly execute changes, all remediation requests are routed through a deterministic Safety Engine that validates them against the live Kubernetes API by checking deployment health, replica counts, PodDisruptionBudget (PDB) constraints, and other availability requirements before any operation is performed. Unsafe actions are automatically rejected, while approved actions are executed and recorded in a comprehensive audit trail for complete transparency. The MCP server exposes Kubernetes functionality through standardized Tools, Resources, and Prompts, allowing AI systems to inspect cluster state, retrieve health information, and perform safe remediation using a common interface. By combining AI-driven decision-making with deterministic policy enforcement, HelmsMan transforms Kubernetes operations from reactive monitoring into intelligent, policy-aware automation that reduces operational effort, optimizes infrastructure costs, and maintains high application availability.
Enterprise decisions often require multiple teams and AI tools for research, financial analysis, risk assessment, compliance, and reporting. These workflows are fragmented, requiring manual coordination that slows decision-making, creates inconsistent outputs, reduces transparency, and increases operational effort. Most AI assistants generate isolated responses instead of coordinating specialized capabilities into a complete, reliable solution.
MADMAX (Multi-Agent Dynamic Matrix) is an Enterprise AI Workforce Operating System built on Model Context Protocol (MCP) and NitroStack. It converts a high-level business objective into an autonomous execution workflow.
An intelligent Planner dynamically discovers MCP-compatible capabilities, breaks objectives into specialized tasks, orchestrates execution, recovers from failures, verifies results, and combines outputs into an executive-ready report. Mission Control provides real-time visibility into capability discovery, planning, execution, confidence scores, verification, and consensus voting, ensuring transparency, explainability, and auditability.
Unlike fixed-workflow AI systems, MADMAX discovers capabilities at runtime, allowing new MCP services to be added without changing orchestration logic. This makes the platform modular, scalable, and future-ready.
MADMAX automates complex enterprise workflows while reducing manual coordination and improving decision quality. By combining intelligent planning, dynamic capability discovery, resilient orchestration, and result verification, it delivers trusted business outcomes instead of isolated AI responses. It can support finance, compliance, HR, customer support, operations, business intelligence, project management, and strategic planning, while seamlessly expanding as the MCP ecosystem grows.
CineOS is an MCP-powered AI Production Manager that transforms ChatGPT into an intelligent filmmaking assistant capable of managing real production workflows. Instead of acting as a chatbot, CineOS uses specialized MCP tools to automate production planning and maintain persistent project data.
The workflow begins with Screenplay Reader, which analyzes a screenplay and extracts key production information. Using this data, CineOS generates optimized shooting schedules by considering actor and location availability. Schedules are stored in a database and can be retrieved at any time using the Get Schedule tool.
CineOS also simplifies production budgeting. Users can calculate and store project budgets, retrieve the budget for a specific film, or list all projects and their associated budgets through dedicated MCP tools. To help productions adapt to real-world conditions, CineOS integrates live weather data. After generating a schedule, it checks the forecast for each shoot day and, if adverse weather is expected, recommends alternative shooting dates or locations to minimize delays and additional costs.
By combining screenplay analysis, intelligent scheduling, budget management, persistent project storage, and weather-aware planning into a single MCP-powered workflow, CineOS eliminates the need to juggle multiple disconnected tools. It enables filmmakers to make faster, data-driven decisions while reducing manual effort and production risks. Designed for independent filmmakers, student creators, and production teams, CineOS demonstrates how MCP can extend AI beyond conversation into a practical production management platform that keeps film projects organized, efficient, and on schedule.
It uses ai algorithms to sort ut most probable stocks for better invvesting.Its specifically made for beginers so they dont loose there money ... Our project maintains strict standard for maximum efficiency.
The problem: India's NBFCs lose qualified borrowers to slow, opaque, consent-blind loan origination. KYC, fraud screening, credit-bureau checks, bank-statement analysis, affordability math, and underwriting are stitched together by call-centre agents and brittle scripts—with DPDP consent treated as a checkbox, not an enforced control.
Vitta is an MCP-native NBFC loan-origination server built on NitroStack and deployed to NitroCloud. It exposes the "hi → signed sanction letter" journey as MCP primitives, letting any client (NitroChat, Claude, ChatGPT Apps) act as the reasoning agent while Vitta is the deterministic capability layer—qualify_lead, record_consent, verify_kyc, screen_fraud, pull_bureau, fetch_bank_statements, compute_affordability, underwrite, generate_offers, create_sanction_letter, and audit, orchestrated tool by tool.
Vitta ships 16 Tools, 5 Resources, and 5 Prompts localized in English, Hindi, and Malayalam. Its engine is a pure deterministic scorecard—FOIR-based underwriting, reducing-balance EMI, hard-negative overrides—using rules, not black-box ML, so every decision is explainable via reason codes, alongside live ECB/Frankfurter FX rates and redacted, append-only audit trails.
The consent gate is the core innovation: pull_bureau and fetch_bank_statements refuse to run without a scoped, time-boxed, HMAC-signed token, making DPDP compliance executable in code. A What-If simulator re-runs underwriting on hypothetical changes—like closing a ₹25,000 EMI—without touching the real case.
Every tool is a drop-in seam for real bureau, Account Aggregator, and e-sign APIs. Being MCP-native, one server serves many AI assistants at once—delivering faster, auditable, consent-first origination for NBFCs and provable compliance for regulators.
MCP Server which allows for direct 3D-game asset generation through blender. The proposed project is an MCP (Model Context Protocol) server that enables AI models to generate, modify, and manage 3D game assets directly within Blender. The main goal of this project is to simplify the game development process by allowing developers, designers, and artists to create high-quality 3D models using natural language instructions instead of manually designing every asset from scratch.
Trade Intelligence MCP is an AI-powered Model Context Protocol (MCP) server that assists banks and financial institutions in automating trade finance compliance, document verification, and fraud investigation. Instead of manually reviewing multiple trade documents, the system orchestrates specialized MCP tools to analyze, validate, and reason across the entire trade transaction. The platform supports common trade finance documents such as Letters of Credit (LCs), Commercial Invoices, Purchase.
Scholar Relay is an AI-powered research assistant built using NitroStack MCP and LangGraph that helps students streamline the research process. It enables users to discover relevant research papers, generate summaries, build a structured bibliography, and seamlessly continue their research across devices using portable MCP Resources. The system also assists in checking citations and identifying unsupported claims while writing, making academic research faster, more organized, and more reliable.
AI Workplace Digital Twin is an AI-powered enterprise assistant that continuously monitors workplace platforms such as GitHub, Jira, Slack, Google Calendar, Gmail, Notion, and other business tools to detect incidents, analyze their impact, and recommend the best course of action. Using a multi-agent architecture, each specialized AI agent analyzes a specific business domain (engineering, operations, support, etc.), while an Executive Agent combines these insights to generate a comprehensive company health report, identify root causes, predict business impact, and suggest actionable recommendations. Once approved by the user, an Action Agent automatically executes the required tasks by interacting with external applications through MCP servers, reducing manual effort and enabling faster, smarter decision-making across the organization.
A persistent memory agent MCP server for students to capture and retrieve observations across lab sessions and lectures. Built for the Amrita University MCP Hackathon 2026.
An MCP-native compliance agent for the EU's Carbon Border Adjustment Mechanism (CBAM).
Since January 2026, Indian steel, aluminium, and other CBAM-covered exporters must report verified embedded carbon emissions per shipment to the EU — or get charged a punitive default rate. That data currently lives scattered across shipment records, supplier declarations, and emission-factor tables, making it slow and error-prone to check compliance manually.
CarbonLot fixes this by treating carbon like a traceability problem, not a spreadsheet problem. Ask it "is shipment SHP001 CBAM-ready?" and it looks up the shipment's supplier, checks whether they've reported verified emissions data, falls back transparently to EU default values when they haven't, calculates the total embedded emissions, and drafts a ready-to-use CBAM declaration — all through natural conversation via MCP.
Built for the Manufacturing & Industry 4.0 Track — MCP Hackathon 2026.
Describe your project, and PreMortem writes a short story of how it fails six months from now — then works backward from that failure to give you a ranked list of real risks and what to do about each one. It's based on a proven decision-science trick: imagining failure before you start catches risks that normal brainstorming misses, because it gets past overconfidence. Instead of a boring checklist, you get a vivid "obituary" for your project, a risk radar, and concrete fixes — so you can defuse the real problems before they happen.
MedTriage MCP — AI-Powered Clinical Triage & Care Navigation
MedTriage MCP is an MCP-based healthcare orchestration system that takes a patient from "something's wrong" to safely matched with the right doctor — closing the gap usually filled by guesswork.
Track 1 (Emergency Detection & Routing): Symptoms are checked against red-flag patterns; genuine emergencies bypass normal flow entirely. Non-emergency cases get matched to an available specialist (searching nearby hospitals if needed), with real working hours confirmed, then booking and payment completed — no patient guesswork.
Track 2 (Medication Safety Auditing): Before any prescription, MedTriage cross-checks it against the patient's active medications, allergies, and history using a real interaction dataset, flagging dangerous conflicts with severity and reasoning. It never recommends a drug — only guards against mistakes.
Track 3 (Rare Disease Registry Matching): For unclear cases, it searches verified registries (FDA approvals, orphan drug data, published trials) for existing, documented treatments — never inventing new chemistry. No match means honest referral, not a guess.
A shared rule governs everything: ambiguity defaults to caution, not reassurance — directly responding to a documented 2026 Nature Medicine finding that a popular AI health tool under-triaged over half of true emergencies. MedTriage separates suggesting from guarding, using only real, cited data throughout.
ORIGO is an agentic AI-powered platform designed to simplify and accelerate the research process. It automatically reads raw research content, extracts meaningful insights, identifies gaps, and generates high-level research abstracts
The Drug-Safety-Checker is a smart AI tool that looks at your medicines and instantly warns you if they are fake, mixed dangerously, or if the dose is too high. It is built for everyday people who take multiple prescriptions, as well as doctors who need a quick way to double-check them. Instead of giving you confusing medical jargon, it uses advanced AI to read complex databases and gives you simple, friendly safety warnings you can actually understand. Ultimately, it acts as a digital safety net, catching dangerous medication mistakes before they happen.
Relay is an AI agent that tracks a shipment across ship, rail, road, and air as one journey, and flags the moment a delay will make it miss a connection between modes — then shows the cost and drafts the fix. Delays only get expensive when they break a handoff (a missed vessel or rail slot), triggering detention, demurrage, and late-delivery penalties that cost the industry around $22 billion a year. Existing tools show where a shipment is; none warn you it's about to miss its train. Relay closes that gap.
B2B sales teams spend a significant amount of time on manual prospecting—discovering potential customers, researching companies, evaluating lead quality, and writing personalised outreach emails. These repetitive tasks slow customer acquisition and are difficult to scale. While modern AI assistants can answer questions or summarise information, they typically cannot autonomously execute an end-to-end outbound sales workflow.
Solution
GrowthPilot is an MCP server built with NitroStack that automates the entire outbound sales pipeline through a modular 7-stage agentic workflow:
Planner → Discovery (Google Places) → Validator → Research (Tavily) → Qualification (Gemini) → Draft → Critic
Each stage is exposed as an independent MCP tool, enabling AI clients to compose, reuse, and extend the workflow. The pipeline integrates live business discovery, company research, lead qualification, personalised email generation, and an autonomous critic–draft revision loop that improves email quality before delivery.
Unlike traditional AI assistants that simply answer prompts, GrowthPilot orchestrates specialised MCP tools to complete real business tasks from start to finish.
Technical Highlights
Built with NitroStack MCP
Full MCP Compliance
9 Tools
1 Resource
1 Prompt
Integrates three live external services:
Google Places API
Tavily Search API
Gemini API
* Autonomous Critic–Draft feedback loop for self-improving email generation
* Production-ready safeguards, including session limits, request queuing, rate limiting, and graceful provider fallback
* Modular, reusable, and AI-client agnostic architecture powered by the **Model Context Protocol**
Impact
A single `gp_run_pipeline` call discovers businesses, researches each company, qualifies leads, and generates personalized outreach emails automatically—transforming a workflow
members:
Suraj.p – am.sc.u4cse24245
Ramrithik Chirra – am.sc.u4cse24219
Dharani kumar Reddy - am.sc.u4cse24123
Srinadh - am.sc.u4cse24262
Most "AI DevOps" demos are chatbots wearing a hard hat. Concord actually runs the pipeline — commit, build, cost-check, canary deploy, monitor, loop — with nobody touching it. It watches five clusters across three regions, live security threats, and real department budgets at the same time, and it doesn't just react, it does the math first: what does halting this cluster actually cost, whose contract does it break, whose budget goes negative. If the answer crosses a hard line, it refuses to act — even fully autonomous, it won't drain a team's budget below its safety floor. Everything else, it just handles.
When a human does need to step in, it hands over a real briefing — financial exposure, conflict type, a live AI-generated root-cause chain — with an SLA clock ticking, not a wall of logs to decode.
Every decision, human or machine, gets hash-chained into a tamper-evident audit trail. Built on Nitrostack/MCP, the same brain runs as a live dashboard and as an MCP server any AI client can call — same rules, same receipts, either way.
RickGotchi is a lightweight, interactive, transparent Desktop Mate featuring on-device facial emotion recognition, system-level coding and music monitors, and an AI-driven debugging advisor that ensures you never give up on your code, and never run around and desert your tasks!
AtlasOS is an intelligent orchestration control plane that transforms natural-language objectives into executable workflows using the Model Context Protocol (MCP). Instead of acting as a conventional chatbot, AtlasOS plans, coordinates, executes, and synthesizes complex multi-step tasks across connected AI agents, enterprise tools, and external services.
A single prompt is automatically decomposed into a structured execution graph, where each node represents a specialized capability such as web research, code generation, database operations, file analysis, API interaction, or business intelligence. AtlasOS dynamically selects the appropriate MCP capabilities, executes them in real time, streams live execution updates, gracefully handles failures through resilient fallback mechanisms, and combines the outputs into a coherent, actionable response.
Designed with a production-grade architecture, AtlasOS features a FastAPI backend, real-time WebSocket synchronization, dynamic workflow visualization, capability discovery, execution monitoring, and fault-tolerant orchestration. Every workflow is generated dynamically—there are no hardcoded execution paths or static responses—providing complete transparency into how AI systems reason, plan, and execute.
Whether automating enterprise operations, accelerating software development, coordinating cybersecurity investigations, or powering intelligent business workflows, AtlasOS serves as the operating system that connects people, AI models, and tools into a unified execution platform.
Key Features
🧠 Natural Language → Executable Workflows
🔗 MCP-Native Tool & Agent Orchestration
📊 Dynamic Workflow Graph Visualization
⚡ Real-Time Execution Monitoring via WebSockets
🛡️ Fault-Tolerant Execution with Intelligent Recovery
🔍 Transparent AI Planning & Decision Making
🔌 Extensible Plugin-Based Capability Registry
🏢 Enterprise-Ready Architecture
VendorIQ is an Agentic AI Procurement Decision Engine designed to help procurement teams make faster, smarter, and more transparent vendor selection decisions.
Today, enterprise procurement is a manual and fragmented process. When a department requests an item (e.g., laptops, servers, networking equipment), procurement teams must gather information from multiple systems such as vendor databases, historical contracts, supplier performance records, compliance documents, and market pricing before selecting a vendor. This process is time-consuming, repetitive, and heavily dependent on the experience of procurement professionals.
VendorIQ transforms this workflow using a team of specialized AI agents. A user simply describes their procurement requirement in natural language, and the agents autonomously understand the request, identify suitable vendors, evaluate vendor performance, analyze historical contracts, generate negotiation strategies, rank vendors using explainable decision scoring, and recommend the most suitable procurement strategy.
Unlike traditional procurement software that only stores information or displays dashboards, VendorIQ actively reasons across enterprise data to support decision-making. Every recommendation is accompanied by an Explainability Tree, allowing users to understand exactly why a vendor was selected based on factors such as cost, delivery performance, quality, compliance, historical relationships, and risk. Additionally, the What-If Simulation Agent enables procurement managers to compare alternative vendors and instantly visualize the impact on cost, delivery timelines, warranty, and overall risk before making a final decision.
The platform is intended for procurement teams, supply chain managers, and enterprise purchasing departments seeking to reduce procurement time, improve vendor selection quality, and make AI-assisted procurement decisions with confidence.
What makes VendorIQ unique is that it does not function as a chatbot
NovaForge is an AI-powered Industry 4.0 platform that uses an MCP server to orchestrate IoT sensor data, historical maintenance knowledge (RAG), and intelligent automation so factories can predict failures, explain why they're happening, recommend the best action, and coordinate preventive maintenance before production stops.
TenderWatch AI is an MCP-powered autonomous agent that analyzes public procurement data, detects suspicious tender patterns, and generates evidence-backed RTI drafts to improve transparency and accountability.
Every factory has the same fight: Production wants the machine running, Maintenance wants it fixed before it breaks. Today that gets settled in hallway arguments while risk keeps climbing.
Downtime Arbiter is a multi-agent system that settles it properly. A Maintenance agent argues from real sensor and risk data. A Production agent argues from deadlines and cost but it only ever sees a coarse urgency level, never the raw numbers. That's what makes it a real negotiation instead of one model talking to itself.
Maintenance's case isn't a guess. It uses causal reasoning based on a P-F curve, a known reliability engineering pattern for how failures escalate once a warning sign appears. So a 24-hour delay and a 72-hour delay can carry very different risk, and the system knows why.
The two agents plan across a rolling two-week schedule, so one decision can affect what other machines are allowed to do too.
When they can't agree, a rule-based Arbiter decides based on real cost, or escalates to a human if it's too risky to decide alone. Every step gets logged, so there's a full record of why each call was made.
Built on the Nitrostack SDK, verified through Nitrostack's Test Cases, and deployed end-to-end on NitroCloud. External components:
Zod for schema validation
CWRU Bearing Data Center for the bearing-spall signal
Groq + Qwen for agent reasoning.
What it does: Analyzes song lyrics section by section, checking syllable consistency, word density, rhyme scheme (including slant rhymes), stress-pattern regularity, and breathing room for a singer — then has the AI actively rewrite any flagged lines while staying true to the song's meaning and tone.
Who it's for: Songwriters, lyricists, and anyone using AI as a creative writing partner who wants to know their lyrics will actually scan and rhyme correctly — not just look good on paper.
What makes it special: Most AI songwriting tools guess at rhythm and rhyme from language patterns alone, since LLMs process text as tokens, not sound. LyricMood is built on the real CMU Pronouncing Dictionary instead — the same dictionary used in speech research — so every syllable count and every rhyme match is a verified fact, not a statistical guess. Even the AI's creative rewrites are held to that same standard: it's instructed to re-check its own suggestions before calling them done.
Impact.OS is an AI-powered multi-agent project management operating system built using NitroStack, TypeScript, and the Model Context Protocol (MCP). It transforms a single natural-language project goal into an autonomous, end-to-end execution workflow through four specialized AI agents collaborating via a shared ProjectState.
Instead of manually managing project planning, budgeting, procurement, scheduling, documentation, reporting, and progress tracking across multiple disconnected tools, users simply provide a high-level objective. The system automatically interprets the goal, generates milestones, creates timelines, identifies campus resources, estimates costs, builds a Bill of Materials (BOM), evaluates sustainability, monitors project health, and produces reports while coordinating all agent interactions through a central Project Director.
Impact.OS follows a human-in-the-loop approach, requiring explicit approval before purchases, resource bookings, meeting confirmations, external communications, or major deadline changes. Deterministic TypeScript services handle calculations, validation, database updates, and state management, while AI focuses on planning, reasoning, recommendations, and report generation. Every workflow is persisted in MongoDB with complete execution logs, enabling projects to resume seamlessly after interruptions.
Designed as a 24-hour hackathon MVP with enterprise scalability in mind, Impact.OS demonstrates how autonomous AI agents can simplify project execution for engineering students, research groups, startups, innovation labs, and campus organizations by reducing administrative overhead and allowing teams to focus on building impactful solutions.
Harmoni-Q is an intelligent DJ assistant built as an MCP server that brings advanced audio analysis directly into any AI chat environment (ChatGPT, Claude, NitroChat).
It solves a core problem for DJs and music enthusiasts: knowing if two tracks will mix well together before ever touching the decks.
Key Features:
1. Deep Audio Analysis: Uses WebAssembly (WASM) based signal processing to extract exact BPM, beat density, and metadata from raw audio files or public URLs.
2. Mix Compatibility Scoring: Compares tracks based on tempo, energy, and rhythm structure to generate a "mixability" score, warning users about potential trainwrecks.
3. Intelligent Set Ordering: Uses graph traversal algorithms and Google's Gemini AI to take a list of disorganized tracks and arrange them into a perfectly flowing setlist that builds energy over time, complete with a narrative explanation of how to execute each transition.
What makes it special is its architecture. By building this as a NitroStack MCP server rather than a standalone app, Harmoni-Q can be instantly plugged into ChatGPT. Users can simply chat naturally with their AI assistant, say "Analyze this track", and Harmoni-Q processes the complex audio physics in the background and visualizes the results with beautiful, interactive React widgets right in the chat.
SentinelMCP is an enterprise-grade Model Context Protocol (MCP) server written in TypeScript using the NitroStack framework. It acts as an autonomous execution and monitoring layer for LLM-driven agents to detect, diagnose, repair, verify, and prevent software and hardware failures on local and remote systems.
Vehicle Genome AI is an industrial-grade engineering platform designed to model, persist, and simulate modular electric vehicle (EV) architectures. Inspired by digital twin methodologies, the application treats a vehicle’s bill of materials as an interconnected "genetic blueprint." It tracks 41 distinct subsystem nodes across powertrain, chassis, thermal systems, and cabin electronics, enabling engineers to synthesize platforms, run compatibility checks, and track lifecycle metrics.
MedSync AI is an MCP application that helps emergency responders, paramedics, and hospital staff coordinate faster during medical crises — from single-patient triage to mass casualty events.
It brings together five coordinated tools: a Mass Casualty Simulator that models resource and hospital allocation for multi-casualty incidents; a Quick Triage Checker that classifies patient severity and suggests the right specialty from reported symptoms; Medication Guidance that surfaces reference-level treatment guidance (with clear clinical disclaimers) for both major emergencies and common ailments like fever or minor cuts; Blood Donor Outreach that matches and notifies eligible, consenting donors near a given location by blood type; and an Emergency Plan Generator that produces a full coordinated response plan for an incident, pulling together ambulance dispatch, hospital capacity, and staffing in one place.
Built on real-time backend services (dispensary, donor, location, hospital, and activity tracking) with a live dashboard, MedSync AI is designed for the moments when minutes matter — giving responders a single coordinated view instead of juggling separate systems for triage, medication reference, blood supply, and dispatch.
It's for EMS teams, hospital emergency departments, and disaster-response coordinators who need fast, reliable decision support without replacing clinical judgment — every medication suggestion carries a clear "reference only, confirm before administering" disclaimer, and donor contact details stay masked until a hospital actually needs to reach them.
# ThreatWeaver – Intelligent Threat Investigation Through AI Reasoning
## What does it do?
ThreatWeaver is an MCP-native threat investigation server enabling Claude to autonomously investigate indicators of compromise across multiple threat intelligence sources, detect coordinated attack patterns, and generate STIX 2.1-compliant reports.
Given an IOC (IP, domain, URL, or hash), ThreatWeaver queries VirusTotal, AbuseIPDB, URLhaus, and Shodan in parallel, correlates findings to detect threat campaigns, maps evidence to MITRE ATT&CK techniques, and generates investigation reports with confidence scores and remediation guidance.
## Who is it for?
Security Operations Center analysts, Incident Response teams, Threat Intelligence analysts, and organizations seeking to augment threat investigation with AI reasoning.
## What makes it special?
**The problem:** Security analysts waste 15-20 minutes per investigation switching between fragmented tools, manually cross-referencing findings, and missing coordinated threat patterns.
**The difference:**
1. **Agentic reasoning** — Claude reasons about multi-source findings and identifies patterns automatically, not just retrieving data
2. **Persistent context** — Investigation state persists across tool calls via MCP Resources, enabling natural follow-up questions without re-prompting
3. **Automatic correlation** — Detects campaigns by identifying shared ASNs, registrants, and malware families across investigation history
4. **MCP-native design** — Demonstrates that MCP enables complex reasoning in specialized domains like cybersecurity, beyond productivity use cases
5. **Enterprise output** — STIX 2.1 reports integrate directly into Splunk, Elastic, and Sentinel
6. **Offline capable** — Works without API keys using pre-loaded threat data
**Result:** Approximately 90 second investigations instead of 15-20 minutes, with pattern detection humans would miss.
TripSync AI is a smart group trip planner that makes travel planning easy and stress-free. Instead of spending hours deciding on destinations, budgets, and activities, users simply enter their preferences, and multiple AI agents work together to create the best travel plan for everyone. Built using NitroStack MCP, the platform uses NitroSDK for AI workflows, NitroStudio for development, NitroCloud for deployment, and NitroChat for interactive trip updates. It helps groups save time, avoid planning conflicts, and enjoy a personalized travel experience.
Blood shortages and delays in finding compatible donors remain critical challenges during medical emergencies. Existing blood donation systems often rely on manual coordination, outdated donor records, and limited communication between hospitals, blood banks, and donors. Our project addresses these issues by creating an AI-powered Smart Blood Donation & Emergency Matching Platform that connects donors, recipients, hospitals, and blood banks through a single intelligent ecosystem.
The platform is designed for hospitals, blood banks, healthcare organizations, voluntary blood donors, and patients who require blood during emergencies. Using AI, the system instantly identifies the most suitable donors based on blood group compatibility, geographical proximity, donor eligibility, previous donation history, health status, and availability. This significantly reduces the time required to locate donors and increases the chances of saving lives.
One of the key features of the platform is its Emergency Request System, where hospitals or authorized users can raise urgent blood requests. The AI prioritizes these requests based on severity and automatically notifies the most eligible nearby donors through the application. Donors can accept or decline requests in real time, allowing hospitals to monitor the status of each emergency request.
The system also includes an AI Health Assistant that guides donors through eligibility checks, answers frequently asked questions, reminds them of their next eligible donation date, and provides personalized health and post-donation recommendations. Additionally, the platform maintains a secure digital record of donations, helping blood banks manage inventory more efficiently and predict shortages using AI-driven analytics.
What makes this solution unique is its integration of AI-powered donor matching, real-time emergency coordination, predictive blood inventory analysis, intelligent donor engagement, and secure digital record management
Event Labs is a production-grade, AI-native event operations platform built on the NitroStack MCP Framewor and Convex Cloud. It enables event organizers and student councils to manage event lifecycles, registrations, QR check-in attendance, judge workload balancing, and cryptographically verified digital certificates—all by chatting naturally with AI agents like Claude Desktop.
QuantGuard MCP is an AI-powered market microstructure surveillance and pre-trade risk intelligence firewall designed for institutional trading desks and portfolio managers.
Built on the NitroStack framework, the server acts as an inline security layer that intercepts dangerous market conditions in real-time before orders are executed on the exchange.
Key Features:
1. Multi-Agent Analysis Pipeline: Orchestrates 7 specialized agents (Liquidity, Toxicity, Spoofing, Volatility, Risk, News, and a Chief Strategy Agent) fanning out in parallel.
2. Microstructure Detection: Implements heuristics for detecting spoofing walls, volume-synchronized probability of toxicity (VPIN) alerts, realized volatility spikes, and square-root model slippage estimation.
3. Deterministic Decision Engine: Integrates a strict rule matrix (e.g., VPIN > 0.7 or spoofing blocks execution) with LLM-generated plain-English strategy memos explaining risk verdicts.
4. Interactive Dashboard Widget: Features custom dark-mode React dashboard components rendered inline in MCP clients, displaying toxicity gauges, spoofing alerts, and pre-trade Value-at-Risk limits.
5. Production-Ready: Full live Binance WebSocket simulation replaying normal-to-toxic market conditions, fully deployed to NitroCloud with GitHub CI/CD integration.
What makes it special is its transition of LLM tools from offline advice to a proactive, inline market guardian. It bridges high-throughput quantitative metrics with natural-language strategic reasoning on any MCP-compatible AI client.
SecretSentry.ai is an AI-powered security scanning tool that analyzes your code repositories for hardcoded secrets, API keys, passwords, and sensitive credentials before they get exposed.
Simply upload your project as a ZIP file, and our NVIDIA NIM agent autonomously reads through your source files, .env configs, and scripts to detect hidden vulnerabilities — giving you a scored security report with exact file locations and remediation steps.
Think of it as a smart code auditor that works in seconds, helping developers catch dangerous leaks before they reach production or get pushed to GitHub.
An AI-Powered Factory Management Agent is an intelligent digital assistant that transforms factory data into proactive insights, enabling manufacturers to monitor operations, predict problems, optimize production, and make smarter decisions in real time.
StudyMate MCP is an AI-powered learning assistant built using the Model Context Protocol (MCP) and NitroStack. The project helps students learn more effectively by exposing educational tools through an MCP server. It provides features such as summarizing lecture notes, generating quizzes from study material, and creating flashcards for revision. The server also exposes educational resources such as course notes and reusable prompt templates that guide the AI to explain concepts according to the student's learning level. By combining MCP Tools, Resources, and Prompts with AI, StudyMate MCP demonstrates how intelligent educational assistants can provide personalized and interactive learning experiences while reducing students' study time.
Our AI agent instantly reviews expense reports by checking every claim against company policies, automatically categorizing expenses, flagging issues like missing receipts or over-limit claims, and clearly explaining why something was flagged. Built for small and mid-sized businesses that still rely on spreadsheets and emails, it reduces manual effort and speeds up approvals. What makes it unique is that it's an MCP server, allowing the same review engine to integrate seamlessly with existing AI assistants, so finance teams can review reports directly through the tools they already use instead of switching to another platform.
What it does: Frankenstein is an agent that builds other agents. You describe what you need in plain English, for example "make an agent that alerts me when O-negative blood stock runs low," and Frankenstein turns that sentence into a real, working AI tool. It figures out what data the agent needs, picks a matching template (vitals monitoring, blood-bank tracking, patient note summarizing), fills in the specific details, tests that it actually works, and deploys it live, all within seconds, with no developer writing code by hand.
Who it's for: Hospitals, clinics, and healthcare teams that need many small, specific monitoring tools but don't have the time or dev resources to build each one individually. Think a nurse who wants a quick alert system, an admin who needs an inventory tracker, or a ward that needs a summarizer for patient notes.
What makes it special: Most hackathon projects build one agent. Frankenstein builds a factory for agents. It's recursive, meaning it can generate new tools on demand, live, in front of judges. Every agent it creates gets tested, deployed, tracked on a dashboard, and logged for audit automatically, so it's not just fast, it's safe and accountable too. That mix of speed, self-generation, and built-in oversight is what turns it from a neat demo into something judges actually remember.
Verifai is a Model Context Protocol (MCP) server designed to detect, analyze, and score AI-generated hallucinations. By comparing agent statements against verified reference sources, Verifai computes deterministic trust scores, flags factual contradictions, and provides structured correction reports with citations.
Atlas Sentinel is an MCP server that gives AI assistants real-time situational awareness for factories and supply chains. It continuously monitors earthquakes, severe weather, space weather, and global news to detect operational risks, helping manufacturers prevent downtime and respond faster to disruptions
What is the Amrita University Amritapuri campus NitroStack hackathon?
The Amrita University Amritapuri campus hackathon is a NitroStack × MCP To The Moon buildathon where student teams design, build, and deploy MCP (Model Context Protocol) apps and servers. Every project on this page is a real, submitted MCP application built by a student team and deployed on NitroStack — the full-stack platform for building and shipping agentic AI and MCP apps.
What are MCP apps and MCP servers?
MCP apps and MCP servers are applications built on the Model Context Protocol (MCP) — an open standard that lets AI agents securely connect to tools, data, and APIs. An MCP server exposes tools, resources, and prompts that any MCP-compatible AI agent can call, turning a large language model into an agentic AI system that can take real actions. NitroStack lets you build, deploy, and scale these MCP apps end to end.
How many projects were built at Amrita University Amritapuri campus?
176 MCP projects were submitted at the Amrita University Amritapuri campus hackathon. You can browse each one below, watch the demo, read the write-up, and open the source code and live MCP endpoint.
How can I build my own MCP app?
You can build your own MCP app on NitroStack (nitrostack.ai) — it provides the SDK, cloud deployment, and studio to go from idea to a deployed, agentic AI MCP server. Explore the docs at docs.nitrostack.ai and browse community builds on r/mcptothemoon.