What it is
txScout is a real-time cryptocurrency forensics tool I built in Python. It watches on-chain activity as it happens, scores how risky any given address is, and traces where funds actually end up once they start moving.
What it does
It continuously ingests on-chain activity and labels addresses against OFAC sanctions listings alongside mixer, exchange and known scam or hack intelligence. Every address then gets an explainable risk score: the score always arrives with the reasoning that produced it, so a verdict can be audited rather than taken on faith.
On top of that sits graph-based fund-flow tracing, which renders the path money takes as a visual graph rather than a wall of transaction hashes, plus real-time alerting when a watched address moves.
Why it works this way
Most blockchain analytics is expensive and closed. You are handed a risk number without the evidence behind it, which for an investigator is exactly backwards: the evidence is the work. txScout was built so that every conclusion can be walked back to the specific transactions supporting it, and so that a trace can be followed by eye instead of reconstructed by hand.
How I use it
I run independent on-chain investigations across Ethereum and Solana. Done by hand, that means pulling transactions, mapping wallets, and correlating what I find against open sources. txScout exists because the repetitive parts of that work were worth automating.
I use it to trace hacks and exploits, map threat-actor wallets, and flag scam patterns before money moves. It is the tool behind the research I do in my own time, and it gets extended whenever an investigation runs into something it cannot yet answer.
Proving it worked
The test I set myself was to reconstruct a real incident end to end. txScout followed the 2023 Euler Finance exploit from the attacker's wallet all the way through to the Tornado Cash mixer, using only publicly available data. No privileged access, no paid intelligence feeds.
Under the hood
A modular ingestion-to-alerting pipeline with pluggable backends: SQLite or PostgreSQL for storage, NetworkX or Neo4j for the graph.
It's a private tool I build and use for my own investigations, so the source isn't public, but I'm happy to walk through the architecture, the scoring logic or the Euler trace in detail.