Getting started

RobinSight

RobinSight reads Robinhood Chain and answers the question a price chart cannot: who is holding this token, how are they connected, and what does the on-chain evidence say about the risk.

Every number on the site is computed from our own index of the chain — every transfer, every swap, every balance — rather than read off a third-party feed. That matters more here than on a large chain: most tokens on Robinhood Chain trade in Uniswap v4 pools that live inside a single contract, and a token whose only market is a v4 pool often has no third-party chart at all.

What you can do with it

Getting started

Quick start

Scanning a token takes one paste. Everything else on the site is a different way of choosing which token to paste.

1. Open the appGo to the scanner. No account and no wallet are needed to start.
2. Paste a token addressAny 0x… contract on Robinhood Chain, or search by name or ticker. You can also click any card on Signals, Pons or PvP.
3. Read the chart firstIt loads before anything else. The rest of the scan — holders, clusters, the audit, behaviour — fills in behind it, each section showing its own loader until its data lands.
4. Check the Verdict tabA risk score out of 100 with the findings that produced it, the contract audit, and an AI read when one is available.
5. Look at the Holder mapThis is the part a chart cannot give you: who holds the token, which wallets are linked, and what the crowd is made of.

Reading a scan in thirty seconds

• Risk score — one number, with every finding that moved it listed underneath.
• Top-10 control — how much of the supply the largest ten wallets hold.
• In sell zone — how much supply sits at or above the multiple this cohort has historically sold at.
• Liquidity vs market cap — thin liquidity against a large notional cap is the most common way a chart lies.

Getting started

Where the data comes from

RobinSight runs its own archive node and its own index of Robinhood Chain. Almost every figure on the site is computed from that index at request time.

The pipeline

• Node — an archive node follows the chain and answers JSON-RPC.
• Indexer — reads block ranges, decodes the events that carry signal (transfers, swaps, pool creation) and writes them to ClickHouse. Every table is keyed on (block, log_index), so re-reading a range after a crash rewrites the same rows instead of duplicating them.
• Sweeps — background jobs that keep derived tables fresh: prices, contract audits, wallet reputation, smart money, token metadata.
• Screener — scores the token set and serves the boards.
• App — the site you use, which reads all of the above.

Why it is built this way

Third-party feeds list the pools they have discovered. On this chain, Uniswap v4 swaps outnumber v3 and live inside one singleton contract, so pool discovery by address does not work. Holding every swap ourselves means a candle can be one minute wide the moment a pool opens, and each bar can carry the buy/sell split and the count of distinct buyer wallets — none of which an OHLCV feed can express.

Finding tokens

Signals

A live board of tokens in their first hour, scored on what actually moved the launches that came before them.

The board exists because of a study of 43,372 tokens launched over four weeks. What it found: about 8% of launches reached 5× against their first hour in the pool. For those, the peak came a median ten minutes after the pool opened, 70% of peaks landed within the hour, and the peak was followed by a 97% fall, with only 6% ever regaining half.

The consequence is uncomfortable and worth stating plainly: a board that waits for a token to prove itself is structurally late. The only moment a signal can be early is the first hour, which is the window this board covers.

The runway

The canvas at the top plots every live launch. Dot size is buyers in the first ten minutes; dots drift right as a launch ages; the left lane holds tokens still on the Pons bonding curve, before a pool exists. Click a dot to find its card, or a card to find its dot.

The four tiers

Tiers are about confidence, not price targets. They say how much the evidence supports the read, not how far a token will go.

Finding tokens

Pons launchpad

Tokens still climbing their bonding curve, before any DEX pool exists.

Pons is the dominant launchpad on Robinhood Chain — around 45% of active tokens come from it. A token on the curve trades against the curve contract itself rather than a pool, so it has no price chart in the usual sense and no liquidity to measure. What it does have is a graduation target, and buyers.

Each card shows progress to graduation, market cap, age, and when it was last bought. Sort by recent buys, newest, market cap, or nearest graduation.

Scanning a curve token

Curve tokens scan like any other token — holders, clusters and behaviour all come from the index. Two things change:

• The plain Chart tab is hidden, because there is no pool to draw. Pro Chart reads the curve's own feed instead.
• The risk score and contract audit are switched off until graduation. While the curve contract owns the market, the questions the audit asks — can liquidity be pulled, can the owner mint — do not have meaningful answers yet.

Finding tokens

PvP — one ticker, every token

Popular symbols are worn by thousands of contracts. PvP answers "which one is the real one" the way an explorer cannot.

Search a ticker and you get every contract using it, ranked by unique wallets that bought from a pool in the last 24 hours, with the previous 24 hours alongside for direction. Around 1,900 tokens call themselves PONS; 1,557 call themselves CASHCAT.

Why buyers, not volume

Traction here is deliberately not a score. Unique buyers is a number you can check against the chain yourself. Volume can be manufactured by one wallet cycling; distinct buyers is harder to fake and easier to verify.

Age is shown for a reason. A contract minted hours ago wearing an established ticker is the copycat tell.

Filters

• Match on ticker or name.
• Exact match, contains, starts with, or ends with.

Scanning a token

The scan page

One token, seven tabs. The header carries the numbers you check first; each tab answers one question in depth.

The header strip

Progressive loading

A scan no longer waits behind a full-page loader. The chart is ready first by a wide margin, so the page opens on it and every other pass lands in the background. A tab whose data has not arrived shows its own loader and a pulsing dot, naming the pass that is still running — "Tracing transfers between holders — page 3 of 6".

Scanning a token

Verdict

Three views of the same question, behind one switch: Risk, AI read, and Contract safety.

Risk

A score out of 100 with a letter grade, and — this is the part that matters — every finding that moved it, each with the metric that produced it. A score with no visible reasoning is a number you cannot argue with; these are all traceable.

AI read

A written read of the scan, generated from the computed metrics rather than from free-form browsing. Every claim cites the metric behind it. It is optional: the site works fully without it, and the panel says whether the read is cached or fresh, which model produced it, and its confidence.

Contract safety

Our own on-chain audit — see Contract safety for what it checks.

Scanning a token

Chart & Pro Chart

Two charts over the same candles. One is an ordinary screener chart; the other draws the wallet layer on top.

Chart

The default, and a conventional screener chart in every respect: candles, a volume strip, timeframes from 1m to 1d, candles or line, log scale, a crosshair legend, a pool selector, and a stat row underneath. Nothing of ours is drawn on it.

Two conventions worth knowing:

• Buckets with no trades are not drawn. The candles pack together and the time axis skips dead stretches, the way every screener does it. The axis is therefore trade-time, not wall-clock — a six-hour silence takes no width.
• Each bar opens where the last one closed. That is the standard charting convention and it is what makes candles join into a continuous ribbon. High, low, close, volume and the buy/sell split are all exactly what traded.

Pro Chart

The same candles carrying everything we know about the wallets inside them:

• Entry and exit markers for profiled holders.
• The median holder entry as a line, so you can see whether the cohort is above or under water.
• The sell zone — a shaded band where this cohort's supply reaches the multiple it has historically sold at.
• An insight strip beneath, each line a computed metric that focuses the chart on its evidence when clicked.
• A per-bar panel — hover any candle for who bought and sold inside it, with buy/sell filtering.

Scanning a token

Trades

The tape under the chart: one row per swap, newest first, refreshing as trades land.

The candles say what a window did; this says what each person did inside it. Columns are time, side, USD, token amount, quote amount, price, trader and a link to the transaction. The bar behind each row is scaled to the largest trade on screen.

The trader column is resolved, not read

A pool's counterparty on this chain is a router or an account-abstraction bundler roughly two times in five. Taking it at face value would fill the column with infrastructure addresses instead of people.

Instead, the wallet is resolved by following the token's own transfer legs and up to four hops inside the same transaction. Addresses reached that way are marked with a dotted underline and say so on hover — so you can tell a resolved trader from one read directly off the swap.

Scanning a token

Holder map

The token at the centre, its holders around it, and every link the index can prove between them.

Bubble size is share of supply. Colour is what kind of wallet it is. Lines are relationships — shared funding sources, transfers between holders, tokens held in common. Drag a bubble to move it, drag the background to pan, scroll to zoom, click for detail.

Colours

Reputation wins over cluster colour: a repeat operator is red even if it also sits in a cluster.

Composition panel

The panel on the left breaks the map down by wallet kind — share of wallets drawn and share of supply, side by side. Both are shown because they diverge, and the gap is usually the point: a cohort that is 4% of the wallets and 60% of the supply is a different token from one where those match. It collapses if you want the map back.

Every legend entry is also a switch — click it to hide that kind of node. The composition panel deliberately does not change with those filters; it describes the token, not the current view.

Scanning a token

Behavior

What the holders have actually done, rather than what they hold.

Each profiled holder gets an entry point, a current multiple, what they have already sold and at what multiple. Aggregated, that answers questions a holder list cannot:

• Is the median holder above or under water?
• How much supply sits at or above the multiple this cohort has sold at before?
• Are wallets adding, trimming, or leaving entirely?
• How much of the current holder base is new demand versus rotation?

Profiling reads each holder's own history, so it is the slowest pass in a scan and the last to finish. The tab shows its own loader until it does.

How scoring works

The seven signals

Every token is scored on seven dimensions. Each is computed from the index and each can be checked against the chain.

Evidence sufficiency

Alongside the score, every token carries an evidence chip saying how much is actually known. "Thin" means strong-looking numbers computed on very little data. A high score on thin evidence is not the same claim as a high score on a full history, and the board does not pretend otherwise.

Collapse memory

Scores do not climb back up after a dump as if nothing happened. Drawdown from a 7-day high, net exodus of holders, outflow, and presence in the rug registry all cap what a token can score, on the board and on the scan page.

How scoring works

Contract safety

Our own audit, computed from live chain state at scan time. Every finding carries the evidence that produced it.

In order of confidence, the audit reads:

1. Verified source and ABI — the precise function list with state mutability, where the contract is verified.
2. Runtime bytecode — when it is not, function selectors are recovered directly from the deployed code.
3. Live state — eth_call reads of owner, paused flags, fees, limits and the like.
4. A simulated transfer — from real holders, to see whether a transfer would actually succeed.

Verdicts are also swept across the indexed token set every few hours so the board can use safety as one of its seven signals without auditing on demand.

How scoring works

Wallet reputation

Rugs are defined by price, not intent. Wallets are judged by how much of their history lives in them.

What counts as a rug

A token that rose sharply from its first traded price, then fell 90% or more from that peak, and has been quiet since. That is a price definition and carries no claim about anyone's intent.

How a wallet is coloured

• Yellow — mostly-rug history. The share of this wallet's traded tokens that later collapsed is high. This is concentration, not count: busy traders touch hundreds of tokens and a few always die.
• Red — repeat operator. A yellow wallet that also bought in the first hour and sold at a profit before the collapse, on at least two such tokens.
• Blue — fresh. Born within a day of buying here, with almost no history of its own.

Airdrops never count toward a wallet's record — a trade has to involve a swap.

Manufactured crowd

When too much of a token's buying comes from mostly-rug wallets — real tokens run under 1% — the crowd is called manufactured, and that caps the token's tier no matter how strong the other signals look. A top-10 holder that is a repeat operator is called out on its own.

How scoring works

Track record

Every call the board makes is written down before the outcome is known, and never edited afterwards.

A "call" is the board moving a token into a tier. The row records the tier and the price our own index showed at that moment. From then on it is only ever marked to market.

Two details make the record honest rather than decorative:

• Every tier is recorded, including the ones we tell you to avoid. Avoid is the control group. If the tokens the board says to avoid do worse than the ones it calls alpha, the signal is doing work. If they do not, the record says so.
• A call has to persist to count. A token has to hold its tier for a dwell period before a call is written, and the same token cannot be called again for a cooldown period. Otherwise a token flickering across a threshold would manufacture a track record out of noise.

Account

Connecting a wallet

Optional. It raises your scan allowance, carries it between browsers, and unlocks deeper tiers if you hold $SIGHT.

How sign-in works

Typing an address into a request proves nothing — anyone can claim any address. So a session is established the way dapps do it:

1. ConnectYour wallet shares its address with the page.
2. The server issues a nonceA one-time value inside a human-readable message.
3. You sign the messageWith personal_sign. This is not a transaction and cannot move funds.
4. The server recovers the signerAnd checks it matches the address being claimed.

Resuming a blocked scan

If you hit your daily allowance and connect a wallet from the notice, the scan you asked for resumes on its own once the signature lands. There is no need to navigate back.

Account

Tiers & $SIGHT

Access scales with how much $SIGHT the connected wallet holds. Balance is read from the chain — there is nothing to buy on the site and no subscription.

Pro also unlocks the two top tiers on the Signals board, AI reads as they switch on, and first access to new tools.

The token

$SIGHT is a token on Robinhood Chain with a supply of 1,000,000,000. It can be scanned in RobinSight like any other token.

Account

Limits

What is metered, what is not, and what happens when you hit a limit.

Scans are metered

A scan is charged once per token per day, before any upstream work begins — so a refused scan is an explanation up front rather than a failure halfway through. Scanning the same token twice in a day costs one scan.

Browsing is free

Signals, Pons, PvP, search, charts, pools and chain stats are not metered. Only the scan-shaped endpoints are.

When you run out

The scan page shows what ran out, when it resets (midnight UTC), and what connecting a wallet would change. If you connect from there, the scan resumes by itself.

API reference

API overview

The endpoints the app itself uses. They are read-only, return JSON, and need no key.

All paths are relative to https://rsight.io. Addresses are lowercase hex with the 0x prefix. Timestamps are Unix seconds.

Errors

Errors come back as JSON with an error key and, where useful, a message. A failing upstream returns a description rather than an empty success — the app shows those strings to the reader.

Signals

The board, tier by tier, capped to the caller's account tier.

Candles

OHLCV from our own swap index, with the buy/sell split and distinct wallet counts per bar — the columns an OHLCV feed cannot provide.

Each candle is an array in the order given by columns.

Trades

Recent swaps, newest first, with the trader resolved through router and bundler hops. Page backwards with the cursor of the last row you received.

Pools & bars

Every pool the token trades in, deepest first, with its quote side, version and swap count.

Everything that happened inside one candle: which wallets bought and sold, how much supply moved, which were new and which left entirely.

Candles composed by wallet action — new, add, trim, exit — by share of tokens moved.

PvP

Every token wearing a ticker, ranked by unique 24h buyers.

Re-reads only the rows on screen. Up to 40 addresses, comma separated.

Audit

The contract audit for one token, computed against live chain state. Each finding carries the evidence that produced it — a selector, a storage slot, an eth_call result, or a simulated transfer.

Reputation

Per-wallet reputation for a token's holders, plus the crowd verdict. This is what colours the holder map.

Related

Tokens this token's holder cohort also holds, computed across their portfolios. Runs server-side across many wallets, so it lands well after first paint.

Also available

The launch board's own feed.

Every token still on a Pons bonding curve.

Account

The caller's tier, today's usage, the tokens scanned today, and the full tier table. Works without a session — an anonymous caller gets the free tier.

Charges one scan credit for a token. Returns 402 with an explanation when the allowance is used.

The sign-in exchange described in Connecting a wallet.