Reading Your Report
This page answers how to read a Spectra report: what the number means, what the dimensions are, and why part of a report can say "unavailable" instead of showing a score.
The Headline Number
The Security Score is a single value from 0 to 100. It is a weighted blend of the dimensions that could be measured for your project. It is not a grade, not a certification and not a prediction of what a project will do next.
Read it together with the breakdown. Two projects can land on the same headline number for completely different reasons, and the breakdown is where that difference lives.
The Dimensions
| Dimension | What it measures | Applies to |
|---|---|---|
| Code | Findings raised by AI analysis against the project's verified source | All projects |
| Sentiment | Community signal on the project | All projects |
| Distribution | How concentrated token holdings are, using measures such as the Gini and Nakamoto coefficients and top-10 supply share | Projects with a token |
| Tokenomics | Supply and token-design signals reported by the on-chain worker | Projects with a token |
| Liquidity | Depth and structure of the token's liquidity | Projects with a token |
| Velocity | Recent trading volume relative to market cap | Projects with a token |
| Holder Growth | Week-over-week change in active addresses | Projects with a token |
| Realized PnL | How balanced holders are between profit and loss. Both extremes score worse than a balanced distribution | Projects with a token |
| Net Flow | Net exchange inflow versus outflow over a 30-day window | Projects with a token |
| Whale Dump Risk | Top-10 concentration combined with how active those holders are | All projects |
Not every audit produces every dimension. That is expected, not a defect.
An Absent Dimension Is Not a Zero
This is the single most misread part of a report.
When a dimension cannot be measured, it is skipped, and the weights of the dimensions that remain are divided by their own sum. The missing one contributes nothing at all, and every other dimension's effective weight goes up to fill the gap. A missing dimension is never scored as 0 and never drags the headline number down.
The practical consequence: on a sparse report the headline score rests on fewer inputs, so read the breakdown before you read the number. Full mechanics are in the Audit Score Reference.
When the Code Score Is Suppressed
The code score is deliberately left out of the headline score whenever the code analysis did not positively complete. It fails closed to absent, never to a passing 100. You will see one of these states:
| State | What it means for you |
|---|---|
ai_unavailable | The AI code analysis did not finish. Nothing was concluded about the code, good or bad. Re-running the audit is the usual fix |
unverified_source | The contract's source code is not publicly verified on the block explorer, so there was nothing readable to analyse. Verify your source, then re-run |
source_fetch_unavailable | We could not retrieve the source. This is usually an upstream or configuration problem on our side, and it is usually temporary |
partial_coverage | Only some of the registered contracts could be analysed. A partial result is not published as a whole-project code score |
reanalyzing | An audit is running right now. The previous code number is withheld while it is recomputed, so you never read a stale score as current |
In every one of these cases the code dimension renders as unavailable, not as a number, and the headline score is computed from the remaining dimensions.
Findings and Recommendations
Findings are the evidence behind the code dimension. Each carries:
- a severity, from critical down to informational,
- a description of the issue,
- the location in the code it was raised against,
- a recommendation for what to do about it.
Severity is what drives the code score. See Finding Severities for the exact penalties and for why the worst core contract, rather than the average across all contracts, sets your code score.
A report with zero findings is not the same as a report where the analysis completed and found nothing. If the analysis did not positively complete, the code score is withheld rather than published as a clean result.
Multi-Token Projects
When a project has more than one token, the distribution dimension blends the per-token scores weighted by market cap, so a large token is not outvoted by a small one. Tokens the worker rejected as invalid addresses do not vote. If two or more tokens share the weight and none of them can be sized by market cap, the distribution dimension publishes no score at all rather than quietly falling back to a plain average.
What the Report Does Not Tell You
- It does not tell you whether to buy, hold or sell.
- It does not certify that the code is free of vulnerabilities.
- It does not cover risk that lives entirely off-chain, such as team conduct or legal exposure.
A high score does not guarantee safety. Results come from AI security analysis with community review, and the product is in beta. Always do your own research.