> ## Documentation Index > Fetch the complete documentation index at: https://docs.codex.io/llms.txt > Use this file to discover all available pages before exploring further. # Token Risk > Read Codex's risk verdict, score, coverage and reason codes for a token or pair, and filter or rank by them. Codex assesses tokens for scam and rug signals and returns one answer: a **verdict**, the **score** behind it, how much **contract analysis** it rests on, and the **reasons** that fired. Use it to warn users and build your own screening. It is a warning signal, not a safety guarantee. ## Where to read it | | Tokens (`token`, `tokens`, `filterTokens`) | Pairs (`filterPairs`) | | - | - | - | | Verdict | `risk.verdict` | `riskVerdict` | | Score | `risk.score` | `riskScore` | | Coverage | `risk.coverage` | `riskCoverage` | | Reasons | `risk.reasons` | `riskReasons` | | Last changed | `risk.flaggedAt` | `riskFlaggedAt` | | Buy tax | `risk.buyTax` | `riskBuyTax` | | Sell tax | `risk.sellTax` | `riskSellTax` | Pair fields describe the pair's token of interest (the one named by `quoteToken`), not the pool itself. A null `risk`, or a null field inside it, means Codex has no assessment. Treat it as unknown, never as safe. ## Verdicts | Verdict | Meaning | | - | - | | `SCAM` | Proven by mechanical evidence (for example, a repeated honeypot simulation or a 100% transfer fee) or labelled a scam by a moderator. | | `HIGH_RISK` | Score of 70 or more. | | `CAUTION` | Score from 40 to 69. | | `NEUTRAL` | Score below 40. Reasons can still be present. | | `VERIFIED` | A moderator marked the token not-a-scam. | | `VERIFIED_CONTESTED` | A moderator marked it not-a-scam, but mechanical evidence disagrees. | `SCAM`, `VERIFIED` and `VERIFIED_CONTESTED` come from proof or a human decision, not the score, so their `score` is usually null. A high score alone never produces `SCAM`. ## Score Each reason adds points, and a 15-point bonus applies when reasons from three or more families fire together. The total is capped at 100. The score is a weighted heuristic, not a probability: 70 does not mean a 70% chance of a scam. Not every reason adds points. Authority reasons (`AUTH_*`), such as a Solana mint or freeze authority, report a capability without scoring it, so you can't rebuild the score by counting reasons. ## Coverage Coverage says how much contract analysis (simulated buys and sells, static checks) the verdict rests on. It doesn't say what the analysis concluded. | Coverage | Meaning | | - | - | | `ANALYZED` | Contract analysis ran and reached a decisive result. | | `INCONCLUSIVE` | Contract analysis ran but was not decisive. | | `NOT_ANALYZED` | No contract analysis. The verdict rests on holder, liquidity, trading and moderation signals only. | | `STALE` | Earlier contract analysis has expired and is waiting to be re-run. | New values can be added to this enum. Treat unknown values as `NOT_ANALYZED`. A `NEUTRAL` token with `NOT_ANALYZED` coverage has had no honeypot or tax check. Show that differently from `NEUTRAL` + `ANALYZED`. ## Reasons [`RiskReasonCode`](https://docs.codex.io/api-reference/enums/riskreasoncode) values are grouped by prefix: | Prefix | What it covers | | - | - | | `PROOF_*` | Mechanical proof, such as a honeypot or non-transferable token. | | `LABEL_*` | Moderator decisions. | | `AUTH_*` | Owner or authority capabilities: mint, freeze, pause, transfer hooks. | | `TAX_*` | High buy or sell tax, or an unconfirmed honeypot result. | | `LIQ_*` | Minimal, unlocked, or suddenly removed liquidity. | | `HOLD_*` | Supply concentrated in the top 10, the developer, snipers, bundlers or insiders. | | `FLOW_*` | Suspicious trading: one-way flow, wallet farms, wash or bundled trading. | | `REP_*` | Creator reputation and imitation of known tokens. | New codes are added over time. Handle unknown codes gracefully, and don't treat an empty list as proof the token is clean. ## Freshness Codex re-assesses tokens in the background as new data arrives; queries read the latest stored result. `flaggedAt` is the Unix time the assessment last **changed**. A re-check that reaches the same result doesn't update it. A new token usually gets its first assessment within a few minutes of its first pool. Until then `risk` is null. Early assessments often have `NOT_ANALYZED` coverage, and the contract analysis fills in later. There is no push when an assessment lands, so poll [`tokens`](https://docs.codex.io/api-reference/queries/tokens) if you need it. ## Examples ### Read one token ```graphql theme={null} query TokenRisk { token(input: { address: "0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2", networkId: 1 }) { symbol risk { verdict score coverage reasons flaggedAt } } } ``` ### Hide risky tokens from a list ```graphql theme={null} query SaferTokens { filterTokens( filters: { riskVerdicts: [NEUTRAL, VERIFIED], riskCoverages: [ANALYZED] } rankings: { attribute: volume24, direction: DESC } limit: 25 ) { results { token { symbol risk { verdict score } } } } } ``` Values inside one filter are ORed; different filters are ANDed. Filtering by verdict also drops unassessed tokens. ### Find likely honeypots among pairs ```graphql theme={null} query HoneypotPairs { filterPairs( filters: { riskReasons: [PROOF_HONEYPOT_SIM, TAX_HONEYPOT_UNCONFIRMED] } rankings: { attribute: riskScore, direction: DESC } limit: 25 ) { results { pair { address networkId } riskVerdict riskScore riskReasons } } } ``` `filterTokens` and `filterPairs` both accept `riskVerdicts`, `riskCoverages`, `riskReasons`, `riskScore` and `riskFlaggedAt` filters, and both can rank by `riskScore` or `riskFlaggedAt`. ## Get the actual tax numbers `risk.buyTax` and `risk.sellTax` return the tax the token charges on each side of a trade, measured by the token's latest decisive contract analysis. The verdict and risk reasons reflect the analysis available at the last risk assessment. On pairs, `riskBuyTax` and `riskSellTax` give the tax of the pair's token of interest. Each side is measured on its own simulated trade: the buy tax on the buy, the sell tax on the sell. Pool fees, Uniswap V4 hook fees and launchpad platform fees are not tax and are left out. A token with a 9% tax on a launchpad that also charges a 1% fee reads `buyTax: 0.09` and `sellTax: 0.09`. A tax-free token in a pool with a 1% hook fee reads `0` where Codex can measure that side. ```graphql theme={null} query TokenTaxes { token(input: { address: "TOKEN_ADDRESS", networkId: 56 }) { symbol risk { verdict coverage buyTax sellTax } } } ``` Reading the values: * They are fractions as numbers: `0.05` is 5%. * `null` means no decisive analysis could measure that side on its own. Treat it as unknown, not as 0%. On some venues only one side can be separated from the venue's fees, for example the buy on Uniswap V3 and V4 pools, so the other side is `null`. * A honeypot returns `sellTax: null` because its test sale failed, so no sell tax was measured. Its `buyTax` is usually set. Read the verdict and reasons for the honeypot itself. * The tax can be newer than `verdict` and risk reasons such as `TAX_SELL_HIGH`. A new analysis shows up at once, while the verdict and reasons update on the token's next assessment. * You can't filter or rank by tax. To find taxed tokens, filter on `riskReasons: [TAX_SELL_HIGH]` or `[TAX_BUY_HIGH]`, which fire at 10% or more. For Solana Token-2022 mints, `extensions.token2022.transferFeeConfig` on [`TokenExtensions`](https://docs.codex.io/api-reference/types/tokenextensions) exposes the configured fee schedules separately from measured risk taxes. A known configured fee can coexist with `risk.buyTax: null` and `risk.sellTax: null` when no decisive simulation established the per-side measurements. The effective fee depends on the active epoch, transfer amount, maximum fee and rounding. Plain SPL tokens have no transfer-fee configuration, so their measured risk taxes remain null. ## How this relates to `isScam` and `potentialScam` Risk filters are opt-in. They don't change the existing scam filters: * `isScam` is a manual label on a small set of tokens. `true` means a moderator marked the token a scam (`LABEL_MANUAL_SCAM`), `false` means a moderator verified it (`VERIFIED`), and `null`, which covers almost every token, means it hasn't been reviewed. Read it from `token.isScam`. The result-level `isScam` on `filterTokens` results is always null. * `filterTokens` hides tokens with `isScam: true` by default. Automated `SCAM` verdicts are not hidden unless you filter on `riskVerdicts`. Pass `includeScams: true` inside `filters` to include them, which you'll need when filtering for `SCAM` verdicts. * `isVerified: true` keeps only tokens with `isScam: false`. It isn't the same as `riskVerdicts: [VERIFIED]`. * `potentialScam` and `potentialScamReasons` are an older automated flag. They're still available, but `risk` is the more complete signal. ### Reading `potentialScamReasons` * The reasons are recomputed from the token's current state each time it is re-scored. A reason drops off as soon as its condition stops holding, and there is no timestamp for when a reason fired. * An empty list means nothing fired. It is also returned when a check was skipped for missing data, so `[]` is not proof the token is clean. * `SniperConcentration` fires when wallets that bought in the first 5 seconds currently hold at least 50% of the supply. It is only computed where sniper stats exist (launchpad tokens). * `MinimumLiquidity` fires when the token's `liquidity` is \$10 or less. `LiquidityUnknown` means liquidity could not be read, so the liquidity checks did not run. * See [`PotentialScamReason`](https://docs.codex.io/api-reference/enums/potentialscamreason) for every value. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.