KRILL turns a messy new token launch into one clear number and a plain-English verdict. No hype, no charts to decode — just a readable signal. Here's exactly how it works.
Every launch gets a single clarity score from 0 to 100. Higher means the launch is easier to read: healthy liquidity, sane holder distribution, and coherent signals. Lower means it's noisy, uneven, or hard to verify.
The score is a weighted blend of the signals below. It's deterministic — the same token always reads the same way, so you can compare launches fairly instead of chasing a number that jumps around.
Readable launch. Clean signals, low friction. Labeled READABLE · decision SIGNAL.
Readable but uneven. One or more signals are weak. Labeled MIXED · decision SCAN.
Hard to read or high risk. Labeled NOISY · decision SKIP.
Five signals carry weight. Holder distribution and contract safety carry the most because they're the hardest to fake and the most likely to hide a rug.
| Signal | Weight | What it reads |
|---|---|---|
| holder_distribution | 40 | Holder count and top-holder concentration. A few whales holding a huge % = riskier. on-chain holder indexer |
| contract_safety | 40 | Honeypot and danger-flag detection — can the token actually be sold. GoPlus |
| contract_integrity | 20 | Ownership, renouncement, and whether the contract's behavior matches its claims. on-chain RPC |
| deployer_reputation | 20 | Who launched this, and how many tokens they've spun up before. on-chain deployer history |
| tax_analysis | 15 | Buy/sell tax, and whether the tax can be changed after launch. GoPlus |
Weights are relative, not percentages — the score normalizes across the signals that were actually measured. A signal with no data is reported as unmeasured rather than scored as zero, so a missing read can never quietly pass as a clean one.
Two hard overrides beat every weight. A confirmed honeypot, or a drain vector like hidden ownership or a ≥50% sell tax, forces NOT SAFE no matter how the rest reads. And if holder distribution or the safety read is missing, the verdict is capped at CAUTION — contract integrity alone never green-lights a token.
| Display-only | Weight | Why |
|---|---|---|
| liquidity_depth | 0 | Read in your browser from DexScreener, which blocks server-side calls. Shown on the card, but never allowed to move a server-authoritative score. |
| social_velocity | 0 | No data source connected. Listed so you can see it isn't secretly influencing the number. |
| narrative_fit | 0 | No data source connected. |
Every scan and every row in the public watchlist carries a safety badge, so you get the call at a glance without reading the number:
The verdict line is written from the strongest and weakest signals, e.g. "Strong liquidity path; watch social velocity. Clean read."
The full launch intelligence — score, breakdown, verdict, and the AI agent — is free for everyone. Holding $KRILL only adds power-user extras. Tier checks are read-only: you paste an address, KRILL reads its balance on-chain. No wallet connection, no signature, nothing to approve.
0x…), or tap an example chip to try one instantly.No. Scores are informational signals about how readable a launch is — not a recommendation to buy or sell. Always do your own research.
No. The score, the full signal breakdown, the AI verdict, and the Ask KRILL agent are all free for everyone — no token, no wallet, no signature. Holding $KRILL only adds power-user extras like priority scans, custom watchlists, and launch alerts.
Never. Tier checks are read-only balance lookups from a pasted address. KRILL never asks for a signature or transaction.
The engine is deterministic by design. A score that flickers every refresh is noise; a stable score lets you compare launches fairly over time.
KRILL runs via Virtuals on Robinhood Chain. On-chain reads (holders, supply, top-holder %) come straight from the chain.
$KRILL is live on Virtuals. Follow @krillintel for launch updates.