On-Chain Due Diligence Workstation

An AI-powered crypto due diligence workstation that turns protocol documents into structured, evidence-backed assessments you can revisit and extend over time.

Overview

This project helps review crypto protocols from the materials analysts already use: whitepapers, audit reports, docs pages, and raw notes. The backend extracts a structured protocol profile with Gemini, scores risk with deterministic rules, stores the result in PostgreSQL, and returns evidence-backed findings to a Streamlit workstation built for repeatable review.

The current implementation is document-driven first. It is not a full autonomous security scanner, and its live on-chain verification is intentionally limited.

Key Features

• Multi-source intake from pasted text, uploaded PDFs, and fetched documentation URLs
• Structured protocol extraction into a consistent project profile
• Deterministic risk scoring across audit status, admin control, tokenomics, and identified vulnerabilities
• Evidence panel that surfaces supporting source snippets for profile claims and findings
• Session-based follow-up chat using saved analysis context and document retrieval
• Saved analysis history with session reload
• PDF export for a completed assessment
• Basic LLM usage tracking for token and cost visibility

Architecture

Frontend

Frontend/ is a Streamlit workstation for: