BFSI & FinTech 404: Sleep Not FoundSubmitted July 26, 2026

TradeMatcher: Autonomous Trade Break Reconciliation Agent

An MCP app on the Model Context Protocol built by 404: Sleep Not Found at the Amrita University Coimbatore NitroStack × MCP To The Moon hackathon and deployed on NitroStack.

About this project

Trade Matcher reconciles trades between two mock trading systems (System A and System B) — detecting matches, price/quantity discrepancies, and unmatched trades — a common and tedious problem in financial back-office operations. Beyond simple matching, it uses an LLM to investigate why a mismatch exists and propose corrections for anything it can't confidently explain, while keeping a human explicitly in the loop for every final decision. load_trades — loads mock trades from System A, System B, or both match_trades — pairs trades by symbol, flags anything outside price/quantity tolerance, or missing entirely from one side investigate_break — an LLM call reasons about why the mismatch might exist (FX timing, settlement windows, or genuinely unexplained) resolve_or_escalate — explained breaks are auto-resolved; unexplained breaks are escalated for human review propose_correction — for escalated breaks only, a second LLM call conservatively proposes a specific fix (which field, which system, what value) with reasoning — but explicitly never applies it run_reconciliation — orchestrates the full pipeline end-to-end in a single tool call and renders results in a live dashboard widget

BFSI & FinTech track

Build AI solutions for banking, payments, insurance, fraud detection, lending, and financial inclusion.

Team 404: Sleep Not Found

  • Ragul PonrajLead

  • Aadidev VS

  • Bavish Nithin

  • Agastya Vuppala

Frequently asked questions

What does TradeMatcher: Autonomous Trade Break Reconciliation Agent do?
Trade Matcher reconciles trades between two mock trading systems (System A and System B) — detecting matches, price/quantity discrepancies, and unmatched trades — a common and tedious problem in financial back-office operations. Beyond simple matching, it uses an LLM to investigate why a mismatch exists and propose corrections for anything it can't confidently explain, while keeping a human explicitly in the loop for every final decision. load_trades — loads mock trades from System A, System B, or both match_trades — pairs trades by symbol, flags anything outside price/quantity tolerance, or missing entirely from one side investigate_break — an LLM call reasons about why the mismatch might exist (FX timing, settlement windows, or genuinely unexplained) resolve_or_escalate — explained breaks are auto-resolved; unexplained breaks are escalated for human review propose_correction — for escalated breaks only, a second LLM call conservatively proposes a specific fix (which field, which system, what value) with reasoning — but explicitly never applies it run_reconciliation — orchestrates the full pipeline end-to-end in a single tool call and renders results in a live dashboard widget
Who built TradeMatcher: Autonomous Trade Break Reconciliation Agent?
TradeMatcher: Autonomous Trade Break Reconciliation Agent was built by team 404: Sleep Not Found at the Amrita University Coimbatore NitroStack × MCP To The Moon hackathon, in the BFSI & FinTech track.
What is an MCP app and how is it built?
An MCP app is an application built on the Model Context Protocol — an open standard that lets AI agents connect to tools, data, and APIs. This project exposes MCP tools and resources that agentic AI systems can call. It was built and deployed on NitroStack, the full-stack platform for shipping MCP apps and servers.