How RektRadar detects scams

Comment RektRadar detecte les scams

RektRadar is a real-time pipeline. It sees every new Ethereum token the moment it is born, runs it through six independent analysis dimensions, maps the network behind it, and keeps watching after launch. Here is exactly what happens under the hood.

Detection starts at the token's birth

Most tools analyse a token after someone asks about it. We start earlier: a watcher subscribes to the creation events of the major decentralised exchanges, so a new pair shows up in our pipeline within a fraction of a second of being created - often before the first buyer arrives. That token is immediately queued for a full analysis.

Six dimensions of analysis

Each token is examined from six independent angles at once. A scam usually has to defeat several of them to slip through - and we surface every signal by name, so the verdict is explainable.

Buy / sell simulation

Before you ever trade, we run a real buy and then a real sell against a forked chain state. If the buy works but the sell reverts, or the effective sell tax is abnormal, the contract is a honeypot or a tax trap. This is the single strongest signal because it tests behaviour, not just code.

Source code patterns

When the contract is verified, we scan its Solidity for dangerous patterns: blacklists, unlimited mint, modifiable fees, hidden owners, drain functions, self-destruct, disguised transfer logic. Verified means readable, not safe - plenty of polished scams are verified.

Bytecode analysis

When a contract is not verified, we read the raw bytecode: dangerous opcodes, suspicious function selectors, and signatures of contracts engineered to behave differently inside a simulator. Reused scam templates are recognised across thousands of tokens.

Liquidity & LP

We check pool depth, whether the liquidity is locked or burned, and who holds the LP tokens. If the creator holds the whole pool unlocked, they can pull the floor in one transaction - the classic rug setup.

Holder distribution

We map how the supply is spread. A token where one wallet holds most of the supply on day one means everyone else is exit liquidity. We also catch sybil distributions - many wallets funded by a single source to fake decentralisation.

Deployer & funding graph

We follow the deployer back through its funding chain and build a graph of who funded whom. This links dozens of separate rug pulls to a single financier - the serial scammers and scam factories that a token-by-token view can never see.

The pipeline, end to end

  1. 1

    Detect

    We watch every new pair and pool the moment it is created on the major DEXs - in real time, straight from the mempool.

  2. 2

    Analyze

    Each token runs through the six dimensions in parallel and gets a 0-100 risk score plus a list of named signals.

  3. 3

    Enrich

    In parallel we crawl the deployer and funder graph and compute a network score and wallet clusters, blacklisting coordinated scam factories.

  4. 4

    Monitor

    We keep watching after launch. Liquidity pulls, fee changes, blacklist additions and ownership changes are caught - some from the mempool, before they even confirm.

  5. 5

    Track over time

    We re-check tokens at intervals to catch slow and soft rugs that only spring the trap days after launch.

All of this turns into a single 0-100 risk score and a list of named signals - 125 of them across ten analyzers, documented one by one in the signal catalogue.

What we catch that a manual check misses

What it does not do (yet)

We would rather be honest than oversell. A clean score means "no obvious scam found", not a guarantee - some traps activate only after launch, and a token that cannot be fully analysed should not be read as safe. No automated tool replaces your own judgement. Use RektRadar to filter out the obvious traps fast, then still do your own diligence on what is left.

FAQ

What is the strongest signal?

The buy/sell simulation. It tests behaviour on a forked chain rather than reading code, so a sell that reverts where a buy succeeded is a structural honeypot with no other explanation. Every other signal is probabilistic.

Does a low risk score mean a token is safe?

No. A low score means nothing obvious was detected. Some traps activate only after launch, and some tokens cannot be fully analysed. Treat a clean result as "not an obvious scam", not as "safe to hold".

How fast is it?

Detection is real time, straight from the mempool. A full analysis runs automatically in about ten seconds - the same checks would take a careful human around ten minutes.

Why so many signals instead of one verdict?

No single check catches every scam. Simulation misses delayed traps, deployer history misses one-off rugs, liquidity says nothing about timed kill switches. We run all of them in layers and surface each named signal so the verdict is explainable, not a black box.

Scan a contract

Go deeper

Signal catalogue - all 125 flags, defined Glossary - market cap, liquidity, slippage and more How to check token safety yourself (manual guide)