These are 58 real MCP apps and servers built and deployed by student teams at the SRMIST 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.
SwiftCare is a multi-agent AI system built on NitroStack that turns a medical emergency report into action in seconds. A Triage Agent analyzes symptoms and vitals to detect life-threatening conditions, a Hospital-Finder Agent locates the nearest suitable hospital, and a Notification Agent instantly alerts the patient's emergency contact with their condition, location, and hospital details — all chained into a single orchestrated pipeline. Built entirely on NitroStack's MCP framework (tools, DI, decorators), tested live in NitroStudio, and deployed on NitroCloud.
Stop reading dashboards. Start asking questions. This NitroStack-powered MCP server connects industrial machine data directly to ChatGPT. By exposing machine telemetry as MCP tools, the AI can instantly fetch and explain the health, temperature, and vibration status of any machine on the floor. We are turning complex Industry 4.0 data into simple, conversational insights, making machine monitoring accessible to everyone.
SkillSync AI – Basic Description
SkillSync AI is an AI-powered peer learning and networking platform built using the Model Context Protocol (MCP) to help students connect, learn, and grow together within their college community. Instead of relying only on online courses or tutorials, SkillSync AI enables students to discover mentors, learning partners, and project teammates based on their skills, interests, and availability.
Students can create professional profiles showcasing their department, semester, skills, projects, and learning goals. The AI Assistant analyzes this information to recommend suitable mentors, generate personalized learning roadmaps, suggest portfolio projects, and build balanced hackathon teams. It also helps students discover peers with similar interests, join learning circles, and share achievements through a collaborative campus feed.
The platform uses modular MCP tools for profile management, mentor matching, networking, learning guidance, portfolio analysis, and social interactions. During the prototype stage, all application data is stored in structured JSON files, making the system lightweight, scalable, and easy to extend.
SkillSync AI aims to transform campus learning by creating a collaborative ecosystem where every student can both teach and learn. By combining artificial intelligence with peer mentorship, it promotes knowledge sharing, teamwork, skill development, and stronger academic communities.
Tagline: Discover. Learn. Teach. Grow.
Project Overview
An end-to-end healthcare solution designed to empower Indian patients by bypassing the traditional general practitioner (GP) layer and routing them directly to the appropriate medical specialists.
Core Features & Architecture
• Context-Aware LLM Chatbots: Utilizes advanced Large Language Models to conduct conversational triage, accurately understanding patient symptoms and medical needs.
• Model Context Protocol (MCP): Integrates MCP layers to efficiently process, manage, and maintain deep user context throughout the patient-bot interaction.
• Secure Information Management: Ensures patient data is safely handled and stored, leveraging Nitro automations in tandem with MCP for seamless, secure backend workflows.
Key Value Proposition
• Streamlined Patient Journey: Reduces wait times and healthcare friction by connecting patients directly with the specialized care they require.
• Intelligent Routing: Employs AI-driven context processing to ensure highly accurate medical referrals.
• Automated & Secure: Combines robust automation and safe storage protocols to maintain data integrity and patient trust.
NovaGear is an autonomous trust gateway for AI-to-business commerce. As AI agents begin transacting on behalf of users — a market McKinsey projects at $3–5 trillion by 2030, already live through ChatGPT Instant Checkout and Mastercard Agent Pay — merchants face a liability they can't see. As Riskified reports, agent-payment protocols pass only minimal data, stripping the fraud signals teams rely on while leaving merchants fully liable: merchants, not LLMs, reimburse banks for chargebacks. Visa's own Trusted Agent Protocol confirms the answer is verifying agent identity in real time.
NovaGear delivers exactly that: a NitroStack-powered agent that orchestrates the full investigation — identity verification, policy enforcement, trust scoring, and settlement reconciliation — through MCP tools. The AI plans and sequences each step, but every security-critical decision runs on deterministic logic, never the model itself — HMAC signature checks, exact payee and integer-amount matches. The result: trust decisions that are accurate, explainable, and hallucination-free before a single transaction settles.
Meeting Supervisor is an AI-powered platform that automates meeting management for remote and hybrid teams. It records meetings, converts speech to transcripts, identifies speakers, and generates summaries, decisions, deadlines, and action items. The system maintains an AI Brain to securely store and retrieve context from previous meetings. It integrates with Google Calendar to manage schedules, identify conflicts, and suggest new meeting times. Team leads can assign tasks while employees can accept, reject, and clarify them. By combining AI analysis, task tracking, and searchable meeting history, it helps teams save time and improve productivity.
India's Ayushman Bharat Health Account (ABHA) under the Ayushman Bharat Digital Mission (ABDM) enables health record sharing, but doctors still manually review fragmented records across providers. MedBridge complements ABHA by reconciling consented records, detecting contradictions, and generating a unified, evidence-backed patient summary for safer clinical decisions.
MedLink is an AI-powered, real-time healthcare ecosystem designed to solve the critical problem of medicine unavailability during health emergencies. Built using Django, Bootstrap 5, Leaflet.js, and FastMCP AI tools, MedLink empowers customers to search medicines by brand or category, view live pharmacy stock bars, and reserve exact quantities for immediate pickup. Pharmacies receive real-time reservation alerts and manage stock via AI automation, while Administrators oversee platform analytics and compliance.
InsightRX is a multi-agent AI system that reads chest X-rays and drafts a clinical report — entirely offline, with a licensed clinician required to approve every result.
What it does: A DICOM chest X-ray plus patient metadata (age, sex, view) enters a LangGraph-orchestrated pipeline. The image is de-identified first — DICOM tag scrubbing and burned-in pixel redaction — and if that fails, the case halts before any AI sees it. A domain-pretrained DenseNet-121 classifies and localizes lung opacities, producing a bounding box and Grad-CAM heatmap so clinicians see where the model is concerned. A diagnosis agent reasons over the findings, an evidence agent retrieves the exact supporting passage from ATS/IDSA guidelines via a signed local FAISS index, and a report agent drafts the write-up. A three-stage verifier firewall then vets it: low-confidence cases abstain, deterministic checks catch inconsistencies, and an independent LLM reviews the reasoning — with calibration (ECE) tracked throughout. Only then does an RBAC-gated clinician approve, edit, or reject, logged to a hash-chained audit trail.
Who it's for: Radiologists and hospitals facing imaging backlogs and burnout — especially resource-constrained or rural facilities without reliable cloud access, who can't risk sending scans to third-party APIs.
What makes it special: Everything runs locally — Llama 3.1 / Qwen2.5 via Ollama, 4-bit quantized — so no patient data leaves the hospital's hardware. PHI is stripped with Presidio. Every diagnosis is grounded in a cited guideline, and every report clears an independent verification firewall before a human signs off. Nothing is fully autonomous, by design.
Future prospects: Monetization follows a compliant per-study SaaS and site-license model, priced against the radiologist backlog it clears, and expands via add-on pathology modules (fracture, nodule, cardiomegaly) and a validated-evidence tier hospitals pay for to support audit and regulatory submissions
Invenio is an autonomous AI research platform that leverages multiple MCP servers and AI agents to automate research discovery, evidence verification, knowledge correlation, and structured report generation from a single query.
An AI-powered project management assistant that helps teams plan projects, generate tasks, track progress, create documentation, and schedule meetings using MCP.Students, hackathon teams, research groups, and anyone working on collaborative projects.Unlike a chatbot that only gives suggestions, ProjectPilot AI takes action. Using MCP, it can interact with tools like Calendar while also generating project plans, tasks, dashboards, and READMEs—all from one platform.
TwinAgent OS is an AI-powered Enterprise Digital Twin that creates a real-time digital representation of an organization's people, projects, workflows, and resources by connecting enterprise tools through the Model Context Protocol (MCP). Rather than acting as a passive chatbot, TwinAgent continuously observes organizational activities, understands relationships between teams and tasks, predicts upcoming risks such as project delays or employee burnout, and proactively executes cross-platform workflows with user approval. By combining AI reasoning, predictive analytics, persistent organizational memory, and explainable automation, TwinAgent transforms disconnected enterprise applications into one intelligent system that helps employees work efficiently, enables managers to make data-driven decisions, and provides executives with a live view of organizational health.
PCOS Monitor is an MCP-powered clinical decision support system that analyzes uploaded hormone reports, compares them with anonymized menstrual cycle patterns and evidence-based PCOS reference data, and generates personalized lifestyle insights, diet recommendations, exercise plans, and an easy-to-understand health summary—all while preserving user privacy by avoiding permanent storage of patient information.
Sentinel AI is an MCP-powered digital evidence intelligence platform designed to assist with the preliminary analysis of digital evidence such as CCTV footage, images, and other digital files.
A unified memory engine that enables every enterprise AI tool—chatbots, copilots, coding agents, and LLMs—to share the same context, ensuring seamless model switching and organization-wide knowledge continuity.
AlphaTex Invoicer MCP transforms static AI chat into a visual, real-time financial command center—replacing clunky text output with dynamic React UI components and laying the groundwork for full-database transactional ERP automation.
Madoff is a production-ready MCP (Model Context Protocol) server that automates end-to-end insurance claim fraud detection using a multi-layer AI pipeline.
When a claim is submitted, Madoff runs a 5-stage investigation pipeline autonomously:
1. Document Retrieval — pulls the claim and its supporting images from Cloudinary
2. Vision AI Analysis — uses Groq's Qwen 3.6 multimodal model to visually inspect document images and cross-check them against the claim text (detecting forged receipts, mismatched license plates, stock photos used as evidence)
3. OCR Extraction — extracts raw text from documents for structured audit trails
4. Rule Engine Scoring — applies deterministic risk rules (duplicate claims, geolocation anomalies, transaction velocity) to generate a 0–1 risk score
5. Final Decision — combines AI confidence with rule scores to auto-approve, reject, or escalate to human review
Built with NitroStack MCP framework, MongoDB Atlas for claim storage, and Groq Cloud for AI inference. Deployed on NitroCloud with both HTTP and STDIO transport — making it compatible with any MCP client including Claude Desktop, Cursor, and Copilot.
The system exposes 11 structured tools including analyze_claim, freeze_account, execute_kyc, check_duplicate_claims, and generate_investigation_report — all callable by any AI agent or human investigator through a unified MCP interface.
Real-world impact: catches fraud patterns that rule-based systems miss, including visually forged documents, cross-border claim inconsistencies, and coordinated fraud rings — all without a single human reviewing the initial claim.
Innovation Gatekeeper is an AI-powered MCP server that automates the technical evaluation of hackathon and innovation submissions. It inspects GitHub repositories, verifies builds, detects originality and licensing issues, evaluates documentation against judging criteria, and intelligently triages projects into review tiers with detailed scorecards and recommendations, enabling judges to focus on the most promising submissions.
DawnMCP is a fully local, privacy-first MCP (Model Context Protocol) server
that gives AI coding assistants deep understanding of a codebase — without
sending any code to the cloud.
WHAT IT DOES
DawnMCP indexes an entire repository locally and answers natural-language
questions with grounded, file-and-line-cited responses — not guesses. It
maintains persistent semantic memory across sessions, so project decisions
and context are remembered instead of lost between conversations. And it
goes beyond Q&A: DawnMCP's AI agents can plan tasks, review code, and help
debug issues by reasoning over the actual indexed codebase.
WHO IT'S FOR
Teams and researchers working with proprietary or unpublished code —
students on unpublished research, companies with sensitive IP, or anyone
who currently can't use cloud-based AI coding tools at all. DawnMCP gives
them a real, capable alternative instead of a memoryless local chatbot.
WHAT MAKES IT SPECIAL
- Zero cloud API calls: every model call runs through a local Ollama
instance (qwen2.5-coder for reasoning, nomic-embed-text for embeddings),
with ChromaDB for local vector storage
- No API keys, no per-token costs — fully offline-capable
- Built on an open standard (MCP), not a closed plugin system — works
with any MCP-compatible client
- Deployable to NitroCloud when you do want a shareable endpoint, while
all inference stays local
Built on the NitroStack framework, DawnMCP also surfaced and fixed a real
dependency-injection timing bug in the underlying framework — where
config-dependent providers could silently fail to initialize under
HTTP/SSE transport — demonstrating genuine engineering depth beyond a
surface-level demo.
Give NitroForge an OpenAPI spec. Get back a verified, working MCP server — typechecked, built, booted, and tested against real protocol calls, in under 15 seconds.Most "AI generates code" tools ask you to trust the model. NitroForge doesn't. We split the pipeline into two fundamentally different kinds of work: judgment and mechanics. The one LLM call in the entire system picks tool names, clusters endpoints, and writes descriptions — things that genuinely require reasoning about what a server should look like. Everything else — the actual input schemas, the field types, the HTTP calls, the validation logic — is derived deterministically from the OpenAPI spec itself and compiled by the real TypeScript compiler. The model literally cannot hallucinate a field name, because it never writes the code that defines one.That split makes NitroForge self-verifying, not self-reported. Every generated server goes through a four-stage machine oracle: real typecheck, real build, real process boot completing a real MCP handshake, and a live replay of every tool against its expected response — diffed, not assumed. If a generated server passes, it's because a compiler and a running process proved it, not because a model said so.
Under the hood, NitroForge is itself a full MCP server — tools for the pipeline, Resources for inspecting what's already been built, Prompts for chaining the workflow into one guided action. It's deployed live on NitroCloud, exposes real, callable tools, and takes real traffic.We built this against a framework with contradictions between its own documentation and its actual behavior — decorator conventions, DI resolution order, transport configuration that silently does nothing unless you know the real mechanism. We found those bugs by reading source, not assuming docs, and fixed several before they became silent production failures. NitroForge isn't a wrapper. It's an argument.
ChiefOS is an AI-powered multi-agent workflow automation platform designed to help organizations intelligently manage emails, meetings, calendars, tasks, approvals, and audit logs from a single unified dashboard.
Built using the NitroStack Model Context Protocol (MCP) framework, ChiefOS employs a Chief AI Orchestrator that analyzes incoming work requests and routes them to specialized AI agents such as Email Triage, Calendar Management, Task Management, and Approval Workflow. Each AI agent performs a dedicated responsibility, enabling faster decision-making, improved productivity, and reduced manual effort.
The platform features a modern React-based dashboard that provides real-time insights into organizational activities, while the MCP backend exposes modular AI tools that can be accessed through NitroStack-compatible clients. This modular architecture makes ChiefOS scalable, maintainable, and easy to extend with additional enterprise AI capabilities.
By combining an intuitive user interface with intelligent AI orchestration, ChiefOS transforms traditional workplace management into an automated, efficient, and AI-driven experience.
What it does: This MCP server gives AI agents a shared, persistent memory. When one agent (e.g., a research agent) discovers something and stores it as a memory, any other agent (e.g., a coding agent) can recall that knowledge instantly — no repeated work.
How it works: Memories are stored in SQLite with 4 types:
fact — something an agent learned
decision — a choice an agent made (with reasoning)
event — something that happened (e.g., handoff between agents)
result — the final output of a completed task
7 MCP Tools exposed:
Tool What it does
remember Store a new memory (fact/decision/event/result) with importance score
recall Search memories by keyword to find what other agents already know
get_task_memory Get all memories for a task, grouped by type
get_decisions Get only the decisions made on a project or task
get_agent_history Get everything a specific agent has done
store_result Store the final output of a completed task
handoff_task Record that one agent is passing work to another
Tech stack: TypeScript, NitroStack framework (@Tool decorators + Zod schemas), sql.js for SQLite storage (pure WebAssembly, no C++ build tools needed).
The code is ready in the repo under nitrostack-server/. The leader needs to either pull it and copy the files from nitrostack-server/src/modules/memory/ into their existing src/ folder, or use nitrostack-server/ as a standalone project.
CareBridge AI is an enterprise-ready, AI-native health triage and patient management platform designed to eliminate critical bottlenecks in healthcare emergency response and clinical intake. By leveraging NitroStack’s Model Context Protocol (MCP) framework, CareBridge AI acts as an intelligent intermediary between patients, diagnostic telemetry, and clinical care teams.
CareBridge AI transforms unstructured patient distress signals (voice notes, multilingual text messages, and vital sensor metrics) into structured, prioritized clinical triage records. Paired with real-time interactive UI widgets, CareBridge AI empowers medical staff to make split-second, data-driven decisions while maintaining strict human-in-the-loop oversight.
BharatFin is an AI-powered financial intelligence platform built on India's Digital Public Infrastructure. Using the RBI Account Aggregator framework, Aadhaar eKYC, UPI, and credit bureau integrations, it securely unifies a user's financial data into one consent-driven dashboard. Powered by NitroStack's MCP architecture, BharatFin automates financial workflows, delivers intelligent insights, and enables faster, smarter, and more transparent financial decision making.
# Ultro
**Touch nothing. Control everything.**
## The problem
Touchscreens assume you can always reach out and tap — not true in an OR, a busy kitchen, a cleanroom, or for anyone with limited hand dexterity. Most gesture demos are flashy toys that spin a 3D object for fun. We wanted gesture control that does real work.
## What it does
Ultro is a touchless control hub: a living plasma orb rendered in WebGL, driven entirely by bare-hand gestures over a webcam. Pinch and move to navigate, two-hand pinch to zoom, open palm to select, point to target a control. Behind the orb sit three panels:
- **Data** — live system metrics you browse without touching a screen
- **Devices** — smart-device toggles fired by gesture, with instant visual + audio confirmation
- **Activity** — a log of every action taken
Every interaction has a distinct sound, not just a visual pulse — so the interface stays usable when you can't watch the screen closely. A full keyboard fallback covers anyone without a webcam.
## How we built it
Three.js renders a 50,000-point additive particle orb with a custom GLSL shader (radial gradient, breathing animation, cursor-driven repel field), composited through multi-pass bloom post-processing. MediaPipe Hands drives real-time landmark tracking with hysteresis-based pinch detection and a small gesture classifier (pinch, palm, fist, point). Web Audio API generates context-aware feedback tied to actions and live data. Next.js powers the app shell.
## Built with
Next.js · Three.js · WebGL/GLSL · MediaPipe Hands · Web Audio API · TypeScript
Contract Sentinel is an AI-powered enterprise risk-assessment and compliance platform that helps legal and operations teams centralize agreements, monitor critical deadlines, and instantly scan contracts for hidden liabilities like unfavorable caps or auto-renewal windows. Designed with a serverless backend and integrated as a custom ChatGPT plugin, the system leverages structured API routes such as `/api/contracts`, `/api/ingest`, and `/api/cycle` to seamlessly ingest agreements, calculate dynamic risk scores, and surface actionable negotiation talking points directly through natural language.
One of the biggest challenges students face during hackathons is forming the right team. Most teams are formed based on friendship or random selection, which often leads to everyone having similar skills while important skills like backend, UI/UX, or DevOps are missing. Beginners also struggle to find where they fit, and organizers spend a lot of time helping students find teammates.
This is the gap our project addresses.
Our solution is the AI Hackathon Team Builder, an AI assistant that helps students form balanced and compatible hackathon teams.
Students first register by entering their skills, interests, experience level, and availability. When a student asks the AI to find a team, the AI uses an MCP server to securely communicate with the backend. The MCP server searches real student data, calculates compatibility, identifies missing skills, assigns roles based on strengths, and even generates a task plan for the team.
As a result, students get balanced teams instead of random ones, organizers save time, and every team starts with clear roles and a proper execution plan.
In short, our project makes hackathon team formation smarter, faster, and more effective by combining AI with real backend tools through MCP.
Thank you.
Vitalis is an MCP server that exposes authenticated tools, resources, prompts, and widgets for clinical information workflows. Vitalis provides clinical decision-support information for research and demonstration. It is not a medical device and must not replace a licensed clinician, emergency services, or local clinical policy.
An Agentic AI platform powered by MCP that verifies enterprise decisions by collecting evidence from multiple sources, detecting contradictions, and generating explainable, trustworthy recommendations with confidence scores.
VeriChain AI is an MCP-powered Agentic AI platform that helps enterprises make trustworthy decisions by automatically collecting evidence from multiple data sources, verifying information, detecting conflicts, and producing explainable recommendations with confidence scores and complete evidence trails.
VidyaAI is an AI-powered personal study companion built on the Model Context Protocol, designed to take a student from "I don't understand this topic" to a complete, structured learning session — research, self-testing, narrated review, and a day-by-day exam plan — all through composable MCP tools rather than a single black-box prompt.
Rampd is an AI agent, built on custom MCP servers, that automates employee onboarding and offboarding across every system a company uses identity/SSO, equipment/asset provisioning, and workspace access (email, Slack, drive) through a single orchestrator layer. Instead of an HR or IT person manually working through 3–4 disconnected tools and risking missed steps (especially on offboarding, where a forgotten access revocation is a real security risk), Rampd composes independent MCP modules into one coherent flow. One command "Onboard Priya Sharma as a Backend Engineer" triggers identity access grants, equipment assignment, and workspace provisioning in the correct sequence, with a single consolidated status returned at the end. Offboarding reverses the sequence, cutting access before revoking identity, closing the security gap that manual processes often leave open.
What it does—
An AI "meeting promise tracker". It listens to meeting transcripts, auto-extracts every commitment ("I'll send the vendor report by Friday"), assigns owners and due dates, then quietly chases them so nothing slips.
Who is it for —
Teams drowning in follow-ups engineering leads and managers who hate chasing people and anyone tired of promises dying after the meeting ends.
What makes it special —
It's fully autonomous follow-through with real Slack nudges on your actual channel, real Linear tickets that auto-escalate to managers with watchers, and it closes the loop by matching Slack/email evidence to mark commitments done. No dashboard to check, no spreadsheet. It runs inside the tools you already use (Slack, Linear, Gmail) like an invisible assistant.
FinPilot AI is a next-generation Model Context Protocol (MCP) server that transforms standard LLMs into autonomous investment bankers. Unlike traditional financial data plugins that simply return raw JSON or text, FinPilot features a Multi-Agent Orchestrator that coordinates specialized sub-agents to fetch live market data, calculate fundamental ratios, run Discounted Cash Flow (DCF) valuations, and assess portfolio diversification.
What truly sets FinPilot apart is its Server-Driven UI architecture. Using NitroStack's widget system, FinPilot dynamically renders stunning, interactive, glassmorphism React micro-frontends directly inside the chat interface. Furthermore, FinPilot bridges the gap between analysis and action by automatically compiling its findings into executive reports and securely emailing them to stakeholders via a custom SMTP integration.
NitroWatch is an MCP server that gives development teams a single control plane for the other MCP servers they're running. Instead of building another agent that performs tasks for an end user, NitroWatch solves the infrastructure problem sitting underneath agentic development itself — the same gaps NitroStack's own team publicly named: too much manual glue code between services, no unified way to monitor deployments, and scattered tooling for logs and token usage.
It exposes four tools and two resources through the official NitroStack TypeScript SDK. register_server adds any MCP server to a watchlist. discover_capabilities connects to that server as a real MCP client — using the official MCP TypeScript SDK's Client over SSE transport — and pulls its actual, live tool/resource/prompt schemas, not assumed or hardcoded ones. get_burn_rate tracks token usage against the hackathon's shared budget. generate_glue auto-generates the connector code needed to chain a tool call on one registered server into a tool call on another, removing the boilerplate that normally has to be hand-written every time two MCP servers need to talk.
The whole system was validated against a real NitroCloud deployment — not mock data — including live introspection of the team's own deployed server and a working connection through ChatGPT's developer mode, confirming genuine MCP-standard interoperability rather than a NitroStack-only proof of concept.
Supply chain disruptions like bad weather, port congestion, strikes, or unexpected events can delay shipments and create costly problems for businesses. RouteGuard AI is a multi-agent AI platform that helps companies stay ahead of these disruptions instead of reacting after they happen. It continuously monitors live risk signals, identifies which shipments and operations are affected, suggests the best alternative routes or carriers, and keeps everyone informed with automatic updates. By helping businesses make faster and smarter logistics decisions, RouteGuard AI reduces delays, lowers operational costs, and makes supply chains more reliable.
SentryFlow is an intelligent return fraud detection engine built on NitroStack by team Peers of Posts (SRMIST) to eliminate "empty-box" return scams on e-commerce platforms like Amazon Seller Central. The second a return package is scanned, SentryFlow analyzes weight discrepancies against catalog specs, computes an explainable fraud score, and gates high-value payouts for human review. Every decision is cryptographically locked in an append-only SHA-256 audit trail and high-risk cases automatically trigger pre-formatted Safe-T claim dispute emails via Resend, protecting seller revenue before refunds are cleared.
Managing a factory isn't easy. Supervisors have to keep track of machines, production, inventory, and maintenance all at the same time. Switching between different systems takes time and makes it easy to miss critical issues. So we built one intelligent platform that brings everything together
To-Do List is a simple and efficient task management application that helps users organize their daily activities. It allows users to create, edit, mark as complete, and delete tasks while keeping track of their progress. The app improves productivity by helping users prioritize important tasks, set deadlines, and stay organized through an intuitive and user-friendly interface.
Vouch is an AI-powered review trust platform that helps users identify genuine reviews by analyzing authenticity, credibility, and supporting evidence. Instead of relying solely on star ratings, Vouch assigns every review a transparent Trust Score using AI, reviewer reputation, verification signals, and community validation—helping consumers make informed decisions while rewarding honest businesses.
Halo is a real-time, biometrically driven pre-ictal seizure forecasting and emergency alerting system. By continuously ingesting multi-modal sensor telemetry from wearable hardware, Halo detects autonomic and kinetic biomarkers of an oncoming seizure—delivering a 10-minute early warning to patients and caregivers before clinical onset.
Aegis is an MCP server that governs other AI agents. Enterprises now connect agents to dozens of MCP tools — file access, email, databases, Slack — approving each one in isolation. Nobody sees the combined result. "Read files" plus "send email" quietly becomes a data-exfiltration path, and this exact pattern is behind real 2026 incidents: the Supabase token leak, the postmark-mcp malicious server that BCC'd every email to an attacker, and Microsoft's tool-poisoning warnings. 88% of organizations reported an AI-agent security incident last year, and Gartner expects most future attacks on agents to exploit access-control gaps.
Aegis closes that gap. It ingests an agent's connected tools, computes its effective permission set (the union of everything it can do across all tools), and detects toxic combinations before they cause harm — then produces an auditable compliance report.
Built entirely on the NitroStack SDK. Tools: connect_tool (OAuth-guarded, simulates adding a tool to an agent), get_capability_graph, detect_attack_paths, apply_policy_fix. Detection is a deterministic graph traversal (no LLM) that finds dangerous capability chains; the policy ruleset is exposed as a Resource and plain-English risk explanations as a Prompt. A React Flow widget renders a live "blast-radius" graph — the agent at center, tools as nodes, dangerous permission-unions glowing red in real time as tools connect.
Who it's for: platform, security, and compliance teams deploying AI agents. What makes it special: it's an MCP that audits the MCP ecosystem itself — turning an invisible, unsolved risk into a single, visual, exportable answer to "what can this agent actually do?"
CottonFlow AI is an AI-powered manufacturing intelligence platform built for the textile industry. It connects spinning, weaving, quality control, maintenance, and production into a unified system that monitors operations in real time, predicts equipment failures, detects anomalies before they become costly, and provides actionable insights through intelligent dashboards. By transforming disconnected factory data into proactive decisions, CottonFlow AI helps mills reduce downtime, improve efficiency, minimize waste, and maximize productivity.
Most defect detection tools stop at "crack detected." That's not useful on a factory floor. You still have to figure out why it cracked, what caused it, whether it's urgent, and what to do about it — and you're doing all of that manually.
CruxAI is our attempt to fix that.
You upload a photo of a component. The system finds the defect, draws a box around exactly where it is, generates a heatmap showing what the model was looking at, and then — this is the part we're proud of — it actually tries to explain what went wrong. Not just a label. It pulls from a vector database of maintenance manuals and past failure records, reasons through the findings using a chain of AI agents, and gives you a root cause (e.g. "material fatigue from uneven cooling during heat treatment") along with specific next steps.
Before any report goes to the engineer, a Verifier Agent checks every claim against the retrieved evidence. If something's not grounded, it gets flagged. Nothing leaves the pipeline unaudited.
The whole thing is packaged as an MCP server on NitroStack — so any AI agent or studio can call into it, trigger inspections, search the knowledge base, or pull machine health history. We also built a full dashboard in React where engineers can upload images, review findings side-by-side with heatmaps, approve or revise reports, and export them as PDFs.
We support three different vision tasks — anomaly detection (MVTec), surface defect classification (NEU, 99.7% accuracy), and steel segmentation (Severstal) — so it works across different line types and materials.
The goal was simple: give the engineer the answer, not just the alert.
LLMs now write most code and get asked to review it for security — but they hallucinate both the bugs and the fixes. TASIE is an MCP server wrapping a ~90-engine detection backend and a Docker exploit sandbox: it finds a vulnerability, sends a real payload at the running app to prove it's exploitable (no proof, no report — which kills false positives), then patches it and re-runs the same attack to confirm the fix holds. Because it's MCP, any client — Claude, ChatGPT, Cursor — calls the same tool and gets ground truth, in the loop where developers already work.
BizShield AI is an all-in-one business intelligence platform built for MSMEs, startups, and local businesses across India. Instead of juggling separate tools for insurance, government schemes, weather risk, and financial tracking, owners enter basic profile details and BizShield instantly generates a Business Health Score, Financial Risk Score, Weather Risk Score, and Market Opportunity Score. It matches businesses to eligible government subsidies and Udyam schemes, recommends the right insurance coverage, tracks compliance deadlines, flags supply chain risks, and delivers real-time weather and disaster alerts with emergency guidance — all through a conversational AI assistant that answers questions instantly, replacing guesswork with data-backed decisions.
In any large organization, decisions are made every day —
vendor rejections, infrastructure migrations, feature
cancellations, project delays. But WHY those decisions were
made? That institutional knowledge lives scattered across
emails, meeting notes, finance reviews, and spreadsheets.
"When someone asks why did we reject Vendor X? —
nobody can answer it without a 30-minute archaeology
expedition through old Slack threads and shared drives."
DecisionTrace AI solves this. It gives anyone in the
organization a single natural-language search bar to instantly
trace and replay any business decision — powered by MCP
Ledger is an MCP-powered platform that turns meeting conversations into actionable commitments. It automatically extracts tasks, assigns ownership, tracks progress and dependencies, and gives teams a clear view of accountability through an interactive dashboard. Built with NitroStack MCP, Ledger helps teams stay organized, follow through, and avoid missed commitments.
“StudyBuddy AI is a multilingual AI tutor that transforms any PDF, PPT, or research paper into interactive explanations, quizzes, flashcards, and personalized revision plans. It helps students learn faster and helps teachers create assessments instantly, making quality education and research support accessible to everyone.”
What does it do?
OrbitGuard is an onboard Model Context Protocol (MCP) server designed for autonomous satellite constellation telemetry classification and fault isolation. Using a multi-stage evaluation pipeline, it:
1. Validates real-time telemetry against critical hardware safety envelopes (voltage, thermal, and tumble rates).
2. Filters out temporary radiation-induced sensor glitches (e.g., Single Event Upsets in the South Atlantic Anomaly).
3. Isolates persistent hardware faults (such as gyroscope drifts) using density-based novelty detection.
4. Generates structured diagnostic reports for ground flight controllers and triggers spacecraft emergency safing modes when necessary.
Who is it for?
It is built for aerospace operators, flight operations controllers, and satellite constellation managers who want to integrate AI copilots (such as ChatGPT, Claude, or custom LLM-based ground control agents) directly into their mission operations for real-time telemetry monitoring and automated troubleshooting.
What makes it special?
- Native MCP Integration: Exposes real-time spacecraft state vectors and safety limits as live MCP resources.
- Space Weather Intelligent Filtering: Distinguishes between transient radiation noise and genuine hardware failures to prevent unnecessary safe-mode entries.
- Copilot-Guided Triage: Bundles structured prompt templates (triage_fault, generate_pass_summary) to give AI agents the exact troubleshooting steps needed to assist human flight controllers during anomaly resolution passes.
The Problem Students learning to code today face a fragmented, overwhelming ecosystem. They suffer from "Tutorial Hell" because they don't have clear roadmaps aligned with real job requirements. They lack access to senior mentors and realistic interview practice. Furthermore, they are disconnected from live tech news and developer communities. Even worse, just starting to code requires hours of downloading compilers, configuring environment variables, and setting up complex workspaces for every new language they want to learn.
How We Solve It Code-To-Career is an all-in-one, AI-powered career accelerator that solves this entire lifecycle in a single platform:
Zero-Setup AI Code Editor: A built-in, AI-powered code reviewer and editor removes environment setup friction entirely. Students can write, review, and debug code immediately in the browser.
Market-Grounded Roadmaps: Instead of generic advice, our autonomous agent searches the live LinkedIn Jobs API for the student's desired skill, extracts real employer requirements, and feeds that to Gemini. The resulting roadmap is grounded in what the industry actually demands.
Context-Aware Mentorship via MCP: Access to mentorship is democratized via our custom Model Context Protocol (MCP) server. It securely exposes the student’s live database state (current roadmaps, milestones, and weak areas) to the AI via a resource:// protocol. The AI mentor literally "reads" their progress before replying, enabling hyper-personalized guidance.
Adaptive Interview Prep: An AI mock-interview agent dynamically generates adaptive MCQs based strictly on the skills in the user's active roadmaps, providing instant feedback.
Community & News: An integrated Q&A community and live-updating Tech News feed keep students connected to the industry.
What Makes it Special? By combining Agentic workflows (taking action before generating content) with novel MCP primitives (exposing database state as context resources), we’ve built an ecosystem that
HealthSync AI is an MCP-powered clinical intelligence platform that orchestrates multiple specialized AI tools to streamline healthcare workflows. Using the Model Context Protocol (MCP), it connects independent servers for FHIR patient records, medical document analysis, drug interaction checks, pharmacy inventory, and clinical notifications. By combining real-time healthcare data with AI reasoning, HealthSync AI helps clinicians make faster, safer, and more informed treatment decisions through a unified, intelligent interface.
DecisionOS is an AI-powered decision intelligence platform that transforms complex business problems into structured, data-driven decisions. Instead of providing a simple chatbot response, it breaks a problem into multiple stages such as context extraction, missing information detection, market research, financial analysis, cost estimation, risk assessment, execution planning, and final recommendations.
Powered by an MCP architecture with 18 specialized AI tools, DecisionOS acts like a virtual strategy team capable of analyzing investments, product launches, business expansion, hiring, operations, and other high-impact decisions. It generates financial insights (ROI, costs, revenue projections), identifies risks, creates execution roadmaps, and delivers a confidence score with actionable recommendations.
Designed for entrepreneurs, startups, students, consultants, and business teams, DecisionOS reduces decision-making time from hours to minutes while improving accuracy through structured AI reasoning rather than generic responses. It provides a transparent, explainable workflow that helps users make smarter, faster, and more confident decisions.
An AI copilot for passport verification officers, built as a NitroStack MCP server. PassportIQ exposes India's 14-stage passport lifecycle as 44 MCP tools and puts an autonomous agent on top: it verifies documents, runs a fraud pipeline, and builds a cross-application link graph to surface fraud rings — then structurally stops at the human officer. Every transition out of officer_review is marked autonomous:false, so the AI cannot decide a citizen's identity. 366 passing assertions.
The SRMIST 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 SRMIST?
58 MCP projects were submitted at the SRMIST 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.