How the Codex score works

The Codex score is an automatic estimate, from 1 to 99, of how risky a token looks. It is the middle of two scores: a Trust score and a Volume score. A higher number means fewer warning signs were found. It is built from public data, so it is an estimate, not an audit and not a guarantee.

Scan a token

Last updated Oct 19, 2026

Three scores

Codex score Main score

The middle of the Trust and Volume scores, rounded to a whole number. The two scores shown next to it always average to it, apart from the cases listed under “What can override the average”.

Trust score

How the token is set up and who holds it: wallet distribution, token age, authorities and taxes, liquidity lock, red flags and other security checks.

Volume score

Whether the trading looks real: liquidity, 24h volume, number of trades, holders, buys against sells, and whether these numbers fit each other.

What the Trust score looks at

Wallet distribution: how much of the supply sits outside the liquidity pools, how much the biggest wallets hold, and whether wallets are linked to each other.
Token age: very new tokens lose points, and the penalty shrinks over the first week.
Authorities and taxes: whether new tokens can still be minted, whether wallets can be frozen, any transfer tax, and whether selling works.
Liquidity lock: how much of the liquidity is locked or burned.
Manipulation signs: for example a honeypot, wash-trading patterns in recent trades, or a copy of a well-known token.
Other security checks: for example a hidden owner, a blacklist, transfers that can be paused, or contract code that is not verified.

What the Volume score looks at

Liquidity: the money in the main pool. Small pools lose points and an empty pool gives the lowest score.
24h volume against liquidity: both very little and far too much trading look unhealthy.
Trades and holders: the number of trades and the number of holders, compared with the size of the pool.
Buys against sells: a token that can be bought but is hardly ever sold is a warning sign.
Fit between the numbers: for example a lot of volume with few trades or few holders, or many holders and almost no trading, can mean bots or fake holders.

How a score is built

Each score starts at 100. Every warning sign takes off points according to a rule table, and each score is kept between 1 and 99. The Codex score is the middle of the two, so the number on the screen is exactly the average of the two numbers next to it.

When a number cannot be found, that costs points too, so the score errs on the safe side. Every deduction is listed in the scan result together with its reason.

What can override the average

A few findings are applied after the average. They are shown as separate lines in the result.

Confirmed red flags set the Codex score and the Trust score to 1, for example a confirmed honeypot, a rugged token, fake liquidity, a copy of a known token or a near-total sell tax.
A pool with no money in it sets the Codex score to 1.
Liquidity that is hardly locked or burned, or a mint or freeze authority that is still open, keeps the Codex score in the Extremely risky band. The more of the liquidity is locked or burned, the higher the best possible score.
A token that is still on its launchpad, before it has moved to a regular pool, cannot score high and has no Volume score yet. Its Codex score follows its Trust score.
A token listed on a major exchange cannot fall below a minimum set by the exchange’s tier: the Very low risk band for the top tier and the Medium risk band for the middle tier. The tier follows a short reference list of well-known exchanges, drawn from public exchange rankings; an exchange that is not on the list counts as low tier, because a listing on an exchange that lists almost every token says little about a token. A lower tier is not a verdict on the exchange itself. A confirmed red flag still sets the score to 1.

Score bands

Codex score	Band
90–99	Very low risk
81–89	Low risk
71–80	Medium risk
55–70	Risky
45–54	Highly risky
1–44	Extremely risky

Trust and Volume scores use the same bands.

Where the data comes from

Prices, market cap and supply for the top 200 cryptocurrencies come from CoinGecko. Pools, liquidity, volume and trades come from GeckoTerminal and DexScreener.

Holders, authorities, taxes and security flags come from sources such as GoPlus, RugCheck, Jupiter, DexPaprika and block explorers, plus a sell simulation from honeypot.is on EVM networks. When two sources report the same figure for the same pool, Codex compares them and marks figures that disagree strongly.

What the score cannot tell you

It is not an audit of the token’s code.
New tokens have little history, so their score can change quickly.
Data can be delayed or incomplete. The result says what was missing.
A token can score well and still lose all its value, and a low score does not prove a scam.
Scores and flags are automatic signals from public data. They are not proof, and they are not an accusation against anyone.

Codex is for information only and is not financial advice.

Quick answers

Does a high Codex score mean a token is safe?

No. It means fewer warning signs were found in the data that was available. The score is an estimate: it can miss risks and it can be wrong.

Why does Codex show two scores and a middle one?

Trust and Volume answer different questions. Trust asks how the token is set up and who holds it. Volume asks whether the trading looks real. A token can look fine on one and weak on the other, so Codex shows both and their middle.

How current is a score?

Each scan reads live data from the sources at that moment, so the score can change from one scan to the next.

Why does a token have no Volume score?

A token that is still on its launchpad has no regular pool yet, so there is no market to measure. Its Codex score follows the Trust score until it moves to a regular pool.