ShipProof
ShipProof is an AI agent that audits submission rules, public links, and code-to-pitch alignment, flagging risks before human approval.
Built at AgentX Hackathon 2026
Team CodeBlooded: Yash Shah, Vedant, Shlok Mane, Tirth J
Built with Python, FastAPI, LangGraph, LangChain, Google Gemini API, Playwright, Pydantic, GitHub API, Gmail API, Next.js, React, TypeScript, Tailwind CSS, WebSockets
About the project
Building a project is only half the challenge. Teams can still lose out because their demo video is private, a repository is inaccessible, submission requirements have changed, or the final product no longer matches what they originally promised.
ShipProof is an agentic submission reliability system designed to catch these problems before they reach the judges.
Powered by Gemini and LangGraph, ShipProof brings submission requirements, project artifacts, and code evidence into a single audit workflow.
It can:
- Read event rules and connected Gmail messages to extract requirements, identify deadlines, and flag conflicting information.
- Check GitHub repositories, demo links, and other submission artifacts from a logged-out visitor's perspective.
- Compare the original project proposal against evidence in the codebase to identify potentially missing features or scope drift.
- Maintain a Coverage Ledger showing which sources were checked, unreachable, or not yet verified.
- Investigate suspicious findings, recheck failures, and prepare suggested fixes, including draft pull requests.
- Pause for human input when requirements are ambiguous and require explicit human approval before an audit can be approved.
Unlike a conventional checklist, ShipProof uses an agentic workflow with tool selection, reflection, project memory, and structured uncertainty handling.
Its core principle is simple: an honest warning is better than a false all-clear.
ShipProof aims to make project submissions more reliable, transparent, and aligned with their original requirements, without taking final control away from the people building them.