SPECTR

Paste a Solana token address, get a risk verdict with the evidence behind it.

Rug pulls are constant on Solana, and most tools hand you a price chart and let you guess. SPECTR returns a verdict (SAFE, CAUTION, HIGH, or RUG) plus the specific flags that produced it.

Tool-calling, not free-form generation

The LLM never invents a risk score. It calls the scoring engine and narrates what comes back: holder concentration, liquidity depth, volume anomalies. Verdicts are reproducible, not hallucinated.

Deterministic rules override the model

Hardcoded thresholds decide the verdict, not the AI. A wallet holding over 80% of supply forces a RUG verdict regardless of what the model concludes — the model explains findings, but never gets the final say on the ones that matter.