Enterprise AI & Workplace Automation Smooth OperatorSubmitted July 26, 2026

Auditor-Zero

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

About this project

Auditor Zero is an autonomous cross-document reasoning agent that catches what human review misses: contradictions between an organization's own documents, and things quietly dropped between versions. Search tools can tell you what a document says right now, they can't tell you what it used to say, why it stopped saying it, or what it silently contradicts elsewhere. Auditor Zero closes that gap. It runs four complementary detectors in parallel: deterministic numeric-conflict checks, LLM-driven semantic comparison, cross-version change detection, and category-disappearance detection, each finding tagged with exactly how it was surfaced. Documents are grouped into version lineages by ingestion order (not fragile version-string parsing), so consecutive versions get diffed for weakened or vanished obligations, while different documents get compared pairwise for contradictions. What sets it apart is the Black Box: every agent decision is sealed into a cryptographically keyed hash chain using HMAC-SHA256, not a plain hash. That distinction matters, a plain hash chain only proves something was logged in order; anyone editing a record can recompute a fresh, internally consistent chain. Auditor Zero's chain is keyed with a server-held secret, so no forged chain can ever reproduce the stored hashes. verify_replay_chain recomputes every link from genesis and reports the exact point where tampering occurred, if any. Every operation, ingestion, contradiction detection, severity scoring, replay, verification is defined once and exposed identically as both an MCP tool and a REST endpoint, so the same reasoning pipeline a human clicks through in the dashboard is exactly what an autonomous agent calls under the hood. Built on NitroStack in 24 hours: TypeScript, Express, the MCP SDK, Claude for reasoning, SQLite, JWT auth, and a full React dashboard with live-streaming audit progress and a tamper-detection demo mode.

Enterprise AI & Workplace Automation track

Develop AI agents and automation tools that improve productivity, streamline workflows, and enhance business operations.

Team Smooth Operator

  • Madhumita ShenbagarajeshLead

  • Hema M

  • Shobhana S

  • Challapalli Sathwik

Frequently asked questions

What does Auditor-Zero do?
Auditor Zero is an autonomous cross-document reasoning agent that catches what human review misses: contradictions between an organization's own documents, and things quietly dropped between versions. Search tools can tell you what a document says right now, they can't tell you what it used to say, why it stopped saying it, or what it silently contradicts elsewhere. Auditor Zero closes that gap. It runs four complementary detectors in parallel: deterministic numeric-conflict checks, LLM-driven semantic comparison, cross-version change detection, and category-disappearance detection, each finding tagged with exactly how it was surfaced. Documents are grouped into version lineages by ingestion order (not fragile version-string parsing), so consecutive versions get diffed for weakened or vanished obligations, while different documents get compared pairwise for contradictions. What sets it apart is the Black Box: every agent decision is sealed into a cryptographically keyed hash chain using HMAC-SHA256, not a plain hash. That distinction matters, a plain hash chain only proves something was logged in order; anyone editing a record can recompute a fresh, internally consistent chain. Auditor Zero's chain is keyed with a server-held secret, so no forged chain can ever reproduce the stored hashes. verify_replay_chain recomputes every link from genesis and reports the exact point where tampering occurred, if any. Every operation, ingestion, contradiction detection, severity scoring, replay, verification is defined once and exposed identically as both an MCP tool and a REST endpoint, so the same reasoning pipeline a human clicks through in the dashboard is exactly what an autonomous agent calls under the hood. Built on NitroStack in 24 hours: TypeScript, Express, the MCP SDK, Claude for reasoning, SQLite, JWT auth, and a full React dashboard with live-streaming audit progress and a tamper-detection demo mode.
Who built Auditor-Zero?
Auditor-Zero was built by team Smooth Operator at the Amrita University Coimbatore NitroStack × MCP To The Moon hackathon, in the Enterprise AI & Workplace Automation 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.