What a token safety score actually measures

A token safety score measures the state of a contract and its liquidity pool at the moment it is read. It checks permissions, taxes, authority flags and pool structure. It does not measure intent, coordination, or whether a token is worth anything — and no score, including the one Meme Central publishes, can.

That gap is the whole subject of this article, because a score's usefulness depends entirely on knowing where it stops.

The specific things a score tests

Strip away the number and the colour, and every safety score in this category is a checklist of on-chain facts. Across providers they cluster into four groups.

Contract permissions. Can supply be increased after launch? Is there an owner, and what can that owner call? Is there a blacklist, a pause, a transfer restriction? Is the source verified? On Solana the equivalents are authority flags: mint authority, freeze authority, metadata update authority.

Trade behaviour. Buy tax, sell tax, and whether a simulated sell succeeds. This is the honeypot test, and it is the check with the highest hit rate and the clearest pass/fail semantics.

Liquidity state. How much liquidity exists, whether the LP tokens are burned or locked, what proportion is committed, and whether the position is removable by the creator.

Holder shape. Concentration in the top holders, and whether the largest holder is the pool itself. Some providers include an insider or bundle heuristic here; the depth varies a lot.

Two providers dominate the infrastructure. GoPlus is a multi-chain security API covering honeypot behaviour, taxes, mint capability and owner privileges — an infrastructure layer that other products embed rather than a place most people visit. RugCheck is Solana-specific contract and LP risk scoring, and it is the standard first check in the Solana workflow. How RugCheck, TokenSniffer and GoPlus compare sets out their coverage side by side.

Why two scores of "90" are not the same thing

There is no standard for token safety scoring. No shared rubric, no shared weighting, no shared definition of what a point is worth. Each provider decides which checks exist, how heavily each one counts, and whether the output is a number, a grade or a list of flags.

The practical consequences are worth stating plainly. A score from one provider cannot be compared to a score from another. A score cannot be compared across chains even within one provider, because the underlying risks are different — there is no mint authority on an EVM chain and no setTax function on Solana. And a numeric score implies a precision that the inputs do not support: the difference between 78 and 84 is a weighting decision made by an engineer, not a measurement of anything.

The more useful way to read any of these tools is to ignore the headline number and read the individual flags. "Sell tax 0%, mint revoked, LP locked, top holder 4%" is information. "Score: 88" is a summary of that information filtered through someone else's opinion about what matters.

What Meme Central's own safety report does — and does not do

Meme Central publishes a per-chain safety report on each token page: GoPlus for EVM chains, RugCheck for Solana. Being specific about its limits is more useful to you than defending it.

It is two different systems, not one. An EVM token and a Solana token on this site are assessed by different providers using different checks and different weightings. The reports sit in the same place in the interface and are not comparable to each other. We surface them that way because the alternative — inventing a unified Meme Central score by normalising two incompatible systems — would look more authoritative and be less true.

It is a snapshot, not monitoring. A report reflects contract and pool state at the time it was generated. Ownership can be transferred, a proxy implementation can be replaced, a lock can expire and a tax can be changed after a clean read. Nothing in a past report constrains a future transaction.

We do not audit anything. We pass through what the provider returns. We have not reviewed the code, we have not verified the team, and a report on this site is not an endorsement, a recommendation or a due-diligence process. It is one automated input.

We do not see the whole market. Meme Central indexes tokens across the venues it covers. That is a sample, not every token on every chain.

A clean report is not a verdict. It means the specific mechanical traps that provider tests for were absent at the time of the check. The overwhelming majority of tokens that lose all their value never fail any of these tests, because going to zero is not a contract defect.

The four things no safety score can see

This list matters more than the checklist above, because these are the categories that produce most losses.

Intent. No scan can tell you the deployer intends to sell their entire allocation on Tuesday. Holding tokens and selling them are not contract flaws — they are ordinary transactions that every scanner will pass. A rug executed entirely through legitimate transfers is invisible to every tool in this category, and that is what most modern rugs are.

Wallet identity and clustering. A holders list shows addresses, not people. Twenty wallets funded from one source in the same block, buying in the same block, look like twenty holders to a scanner reading concentration. Detecting that requires cluster analysis — Bubblemaps is the standard for supply-concentration and wallet-cluster visualisation, and nothing else substitutes for it. Sniper wallets and bundled supply at launch covers the pattern; reading a token's holder distribution covers what the raw numbers can and cannot show.

Everything off-chain. Paid engagement, bought followers, a fabricated team, a fake endorsement, a compromised account. Robinhood Chain produced a clean example on 23 July 2026, when Vlad Tenev's X account was compromised and used to promote a fake "Vladhood" token described as the chain's official mascot; Robinhood confirmed the compromise. A scanner reading that token's contract would have found a contract. The red flags in a memecoin's social presence are a separate pass, and it has to be done by a person.

The deployer's other wallets and other launches. A creator who has deployed and abandoned forty tokens scores identically to a first-time deployer on their forty-first, because the score reads the token, not the history behind it.

How to use a score properly

Treat it as a filter, not a verdict — a way of removing obviously broken tokens quickly so you can spend attention on the ones that survive.

A failing score is highly informative: it means a specific, named, mechanical problem exists, and you should almost always act on that without further investigation. A passing score is weakly informative: it means the tested traps are absent, and the questions that matter most have not been asked yet.

The sequence that follows from this is: automated scan first because it is fast and cheap, then read the contract yourself, then holder and cluster analysis, then the off-chain surface, then a position size that assumes you missed something. The last step is the one that actually protects you.

Frequently asked questions

What does a RugCheck score mean?

It is RugCheck's own weighted summary of Solana-specific checks — mint and freeze authority state, LP status, holder concentration and related contract risks. The weighting is RugCheck's editorial choice, so the number is not comparable to any other provider's. Read the individual flags rather than the total.

Is a token with a good safety score safe to buy?

No. A good score means the mechanical traps that provider tests for were absent when the check ran. It says nothing about the creator's intentions, coordinated wallets, off-chain deception, or demand for the token. Most memecoins that go to zero pass every automated check on the way down.

Why do different tools give the same token different scores?

Because there is no shared standard. Each provider chooses its own checks and its own weightings, and some are chain-specific while others are multi-chain. A disagreement between two tools is usually a difference in scope rather than one of them being wrong. Compare the underlying flags, not the totals.

Does Meme Central's safety report catch everything?

No, and we would rather say so. It passes through GoPlus on EVM chains and RugCheck on Solana, so it inherits their coverage and their blind spots. It is a snapshot rather than continuous monitoring, it is not an audit, it does not detect intent or wallet clustering, and it covers the tokens Meme Central indexes rather than the whole market.

Can a token pass a safety check and still be a rug?

Routinely. Passing means liquidity is committed, permissions look constrained and the sell simulation succeeds. It does not mean the creator's allocation is small, that the top holders are unrelated, or that anyone intends to keep supporting the token. Selling a large held allocation into the pool requires no special permission at all.

---

The check a creator can pass in advance

Every limitation above describes a buyer trying to infer intent from the outside. The only part of that a creator can actually settle is the liquidity question. A time-locked LP position through Team Finance — built by TrustSwap, which also builds Meme Central — locks LP tokens on Ethereum, Robinhood Chain, Polygon, Base and BNB and appears as a verified badge on the token's page in the token feed on Meme Central. It applies to a token from any launchpad, including after launch. It will not improve a contract-permissions flag, and it does not stop you selling your own supply — which is precisely the thing no score can see.

---

Nothing here is financial, legal or tax advice. Memecoins are extremely high-risk: most lose most of their value, and the majority of tokens launched never reach a decentralised exchange at all. Never spend money you cannot afford to lose entirely. Meme Central does not recommend any specific token. Data described as Meme Central's own reflects tokens indexed by Meme Central and is not whole-market data.