Token intelligence for Robinhood Chain

The chart is where the story ends.

A candle tells you what price did. Sherwood Insights reads what it cannot: who holds a token, how those wallets are linked, where they have sold before, and what the contract can still do to you. Every finding carries the evidence that produced it.

Two views of the same token

Same hours. Same price. One of these tells you what happens next.

A price series is the last thing to know about a token, because it is the only thing everyone already knows. The signal is in the wallets underneath it.

What everyone sees7-day close · pool OHLCV
d-7d-3now

Up and to the right. Nothing on this surface says who is buying, or whether the people already in are about to leave.

What Sherwood seessame series · wallet layer on
11 wallets, one funding source shared funder · treated as one entity Owner renounced · minter() live residual authority · evidence: getter Past the cohort's trim multiple these wallets sold here before top-10 · entity-adjusted d-7d-3now

The same line, with the wallets and the contract laid over it. Each note is a computed metric, with the evidence that produced it one hover away.

What a chart cannot compute

Signals that live in the wallets, not the candles.

Each one is computed from the chain at scan time, from transfer history and contract state, never from a vendor snapshot. The number and the evidence travel together.

01 / 06

Hidden connectors

Wallets that never transacted with each other but were funded by the same address are drawn as a dashed link and counted as one entity. The chart sees eleven buyers. The map sees one.

shared funding sourcetransfer graphcluster detection

Entity-adjusted control

Top-10 share counted twice: once by address, once with linked wallets folded into one holder. The gap between the two bars is how much control the raw number was hiding.

top-10 entitygini · hhinakamoto

Sell-pressure read

Up to 300 top holders profiled from their own history: entry price, current multiple, and the multiple at which they trimmed last time. Then the share of supply now sitting past that point.

supply in sell zonemedian ÷ trim multiplewallet classes

Related tokens, with lift

What this token's crowd also holds, ranked by wallet count and paired with lift against the whole chain. On ARROW, DIH tops the list with 73 wallets but only 3.6× lift. NVDA sits lower at 45 wallets and 41.7× lift. The second one is the finding.

lift = cohort share ÷ chain share200-wallet cohort

Market-cap habitat

Where this crowd's positions sit, drawn twice: by count and by dollars. On ARROW the median position is a $565k token, but 43% of their money sits above $250M. ARROW is at the 91st percentile by count and the 17th by dollars. A count-only chart implies the opposite conclusion.

by position · by dollarkeyed by contract, not symbol

A contract audit that shows its work

Admin powers, residual authority, live taxes, proxy slots, self-destruct opcodes and a simulated sell path, all read from the chain. UP scored 90/A under a vendor feed that had no data for it. The local audit scores it 2/F: mintable through a live minter role while ownership is nominally renounced, with top-10 entities holding 97%.

evidence per flageth_call · storage slots · opcodesunknown ≠ safe
0holders mappedBubble-mapped by share of supply, with real transfers as links.
0wallets profiledFrom their own transfer history: entry, multiple, what they sold and when.
0wallet cohortPortfolios read for related tokens and habitat. Pools and routers excluded.
0vendor snapshotsEvery audit finding is read live from the chain, with its evidence attached.

How a scan runs

Six passes. About a minute. Whole when it lifts.

A scan pulls identity, contract state, holders, transfers, price and behaviour before the page is usable. The veil stays blurred rather than opaque, so you watch the panels assemble underneath.

identityToken metadata and the address resolved, keyed by contract so two tokens sharing a ticker never merge.
contractVerified ABI or 4-byte selectors from bytecode, owner and minter getters, proxy slots, opcode walk, simulated sell.
holdersTop holders with a batched code check, so routers and pools are never profiled as people.
transfersThe graph between holders, and the funding sources behind wallets that never touched each other.
pricePool OHLCV for the sparklines that have a real series. Nothing is drawn over data that does not exist.
behaviourPer-wallet entry, multiple and trim history, then the supply now sitting past where this crowd exited before.
identity
contract
holders
transfers
price
behaviour
resolving token…eta —

Evidence, not vibes

A risk read where every sentence cites its metric.

Optionally, Claude narrates the computed metrics. Only the metrics. Each claim points at the id that produced it, and it will not tell you to buy or sell.

Grounded in the scanThe model sees the metrics JSON and nothing else. No web, no priors about the ticker, no vibes.
Unknown stays unknownWhatever the audit could not establish is reported as a hole, not a pass. A false "safe" is worse than an admitted gap.
Refuses to adviseThe read describes risk. The decision stays yours, and the page says so rather than hedging.
AI read · groundedclaude · metrics only

Ownership is nominally renounced owner_renounced but a live minter role remains minter_live, so supply can still be expanded by a party the renounce did not remove.

Concentration is severe once linked wallets are folded together: the top ten entities hold 97% of supply top10_entity, against a much lower raw address figure top10_raw.

Transfers to the pool are not blocked sell_path_clear. Whether this deployer has shipped honeypots before was not assessed deployer_history: unknown.

You should buy / sell.  Not a sentence this model is allowed to produce.

Start with any address

Scan the token before you touch it.

0x paste a token address or search by name Scan a token

Runs locally on public endpoints with no keys. Add an Anthropic key and the AI read turns on.