Trust the evidence.
Question the verdict.
A transparent method is more useful than a mysterious score.
How we score
Hard blockers come first. Active mint authority, active freeze authority, creator-held LP, or a transfer fee above 25% forces Severe risk. Good signals cannot average these away.
Other warnings add weights: linked holders 25; deployer history 25; top-10 concentration 15; transfer fee, deployer holdings, first-block buyers and liquidity ratio 12 each. Scores at 55 or above are Severe, 30 or above High, and positive scores Elevated. Low requires every check to have evidence with no warnings. Missing evidence sets a floor of Elevated.
Thresholds follow the project plan: fee above 5%; at least 8 linked top-20 holders; concentration above 25%; deployer holdings above 5%; majority of prior tokens dying within 24 hours; first-block share above 30%; liquidity below 5% of market cap. Pool age under 24 hours is context only. Thresholds are initial heuristics, not calibrated probabilities.
What the AI can say
AI selects the relevant signal evidence for three analyst perspectives and the consensus. The application renders verified facts and their explanations from those signal IDs. The model cannot introduce numbers or unsupported claims. If AI is unavailable, the panel is explicitly labelled computed evidence.
Limits are part of the result
Risk assessment, not financial advice. Low risk does not mean safe. On-chain data cannot detect abandonment or narrative collapse.
Funding graphs use bounded transaction samples. Shared exchange funding does not prove shared control. First-block recipients can include transfers, not only buyers. Provider holder labels may be incomplete; LP status stays unknown when a reported pool cannot be verified. Prior-launch outcomes stay unknown when attributable evidence is missing. A transfer from a deployer is never presented as a confirmed sale without swap evidence.
Commercial independence
Scores cannot be purchased. There are no paid placements or “verified safe” badges. Alert data is not sold to trading bots. The team must not trade against scanner users or trade tokens flagged by the tool. Any paid promotion must be disclosed.
Launch gates & disclosures
Accuracy gate: pending a labelled historical dataset and documented backtest. Traction gate: pending repeat-use evidence. Demand gate: pending evidence of paid monitoring demand. Token launch remains unapproved. Public browsing and basic scans are enabled; subscriptions and authenticated integrations remain limited by configuration and verification.
Team allocation, team wallets, vesting, access-token mint and LP lock/burn evidence: not supplied. Treasury: BNqu9sdYdFWEmqujfY2YVAdaQPkHeBbH7rWUZEA5P5bc. Legal review: not supplied. Public application: rugsignal.com. Social handles: not confirmed.
Provider usage/resale terms: RugCheck, Helius, DexScreener and AI commercial-use approval must be confirmed before public release. No claims of legal clearance or calibration are made.
Data & privacy
Basic scan reports are public by their share link. Wallet sessions, watchlists, Telegram links and billing records are private. Wallet sign-in only signs a message; it does not request spending authority. Sessions expire after 12 hours. Holding entitlements are checked on each paid operation and alert delivery.
Scans retain immutable evidence summaries for accuracy comparison. Authentication challenges expire after five minutes. Logs exclude credentials and signatures. Operators should apply the documented backup and retention policy.
Inspect the accuracy log ↗