Method and limits
The only thing that makes a risk score worth anything is being able to see how it was produced. This page sets out what the score is made of, what data it rests on, and what it cannot catch.
Starts at 100
Every token enters with 100 points. Each risk signal found subtracts from it; nothing adds back.
A veto zeroes it
For a token that cannot be sold, the other findings are meaningless. If a honeypot is found the score is zero outright.
Unknown is not clean
An item with no data costs half its weight. Missing information is not read in the token's favour.
No known critical risk signal. This is not a guarantee, only the absence of the risks that can be measured.
More than one warning. Read which findings cost points before you go in.
Serious risk signals, or a veto finding outright.
The penalty table was designed for newly launched tokens. An open mint authority and unrenounced ownership on a new token is a serious risk. But the same privileges belong to the MakerDAO protocol in DAI and to Paxos in BUSD; they have sat open for years without being abused. Concentration in the top 10 wallets at that scale is mostly exchange and protocol contracts.
So for tokens that clear measurable thresholds on age, liquidity and holder count, some penalties are reduced. Reduced — not removed. The privilege is still there; it is simply weighted against its history. Every reduction is written out finding by finding in the audit report.
Blue-chip token
All three measures must pass:
Established token
All three measures must pass:
The values are the maximum penalty for that finding; if the finding is merely suspicious (amber), 60% of that figure is deducted instead.
- The honeypot veto — a token that cannot be sold cannot be sold at any age.
- Buy and sell tax — a real cost you pay today, which no history softens.
- Identity and transparency, volume-to-liquidity ratio, pool age, holder count.
In a bundled buy, one person buys the same amount at the same moment through many wallets. The result is that those wallets end up holding almost exactly equal balances. In organic buying that coincidence does not happen: everyone buys at a different time, in a different amount.
RugScan filters locked pools, exchange and contract wallets out of the holder list, then looks among the remaining personal wallets for the largest cluster whose balances sit within 15% of each other.
GeckoTerminal
go to sourceThe new-pool feed for every chain, with token images. This is what the radar runs on.
DEX Screener
go to sourcePrice, liquidity, volume, trade counts, pool age, linked site and social accounts.
GoPlus Security
go to sourceHoneypot and sell simulation, buy/sell tax, mint and ownership privileges, freeze, blacklist, hidden owner, LP lock, holder distribution.
Etherscan & Helius
go to sourceThe developer record: who deployed the contract and what that wallet has launched before.
Bubblemaps
go to sourceA visual map of money flow between wallets. RugScan links to it directly for every token.
All queries are made from the RugScan server; your browser never connects to these services directly and no API key ever reaches the client.
RugScan does not remove risk, it changes the odds. It never says «this token is safe»; it uses the language of known risk signals.
The developer who sells later
A token with a clean contract, locked liquidity and reasonable distribution can still collapse the day its developer sells. There is no on-chain sign of that beforehand.
The score is a snapshot
If ownership has not been renounced, the rules can change five minutes from now. It shows the state at the moment you ran the scan, not afterwards.
The professional with a fresh wallet
A token launched from a brand-new wallet with no history looks clean in the record. That is why a separate «wallet is two days old, no history» warning exists.
The bonding curve phase
On platforms like pump.fun there is no classic LP pool until the token migrates, so the liquidity lock check gives no meaningful result there.
Ready to audit a token?
Back to the audit desk