About CanaryTape

CanaryTape is a free, public scam checker for Solana tokens. Anyone can paste a token and get an explained Canary Score built from public on-chain data, or paste their own wallet into Was I the exit? to see who was selling around their buys. There is no sign-up, no wallet connection, no paid placement and no affiliate link.

Meme-coin launchpads create tens of thousands of tokens a day, and many are drained within hours. The warning signs are usually visible on-chain before the crash: supply bought in the launch block by bundled wallets, holders funded by the same wallet, creators selling, liquidity that can be withdrawn, and live mint or freeze authority. CanaryTape puts those facts in one place in plain language.

What the tools do

• X chatter: for each Top Rugs coin we count posts on X that contain its mint address over the last 7 days (X recent-counts API, refreshed at most once a day), and label the mood of up to 25 recent posts as Positive, Mixed or Negative with a simple public keyword list (fewer than 5 posts: “Not enough posts”). The sample size is shown. We also flag when mentions in the 6 hours before the price peak ran at 3× or more the usual hourly rate. Only aggregate numbers are shown: no post text and no handles. Bots, giveaways and paid promotions inflate counts, a cashtag-only mention is not counted, and chatter is not a safety signal.
• Buzz: a compact per-coin view of where hype shows up beyond X, built only from free, public, keyless sources and stored as numbers only (no post text, no usernames, no group names). Per source: X posts containing the mint (7 days, from X chatter above); DexScreener buy and sell trades across the coin's Solana pairs (24 hours); GeckoTerminal unique buyer and seller wallets (24 hours) and whether the coin is on its trending list; Telegram member count of the public group the coin itself lists, when it lists one. The combined mood is a simple rule-based tally, not a judgement: the X keyword label counts +1 / 0 / −1, and each market source counts +1 when buyers clearly outnumber sellers, −1 when sellers clearly outnumber buyers, 0 otherwise (small samples are skipped). It reads Positive, Mixed or Negative from the average, or “n/a” when no source had enough data. Buy/sell balance is market activity, not opinion. A source that fails shows “n/a”.
• Paid promotion flag: shown when DexScreener's public orders data lists a paid order for the coin (an enhanced token profile, an ad, or a community takeover listing) or paid boosts, or the coin sits on DexScreener's boosted lists now. These are bought placements, so we label them “paid promotion”, never organic interest. The hover text shows what was paid for, how many paid in the last 7 days, and when the last payment was. Paying for promotion is common and is context, not an accusation.
• Canary Score: ten checks, 100 points. Our own on-chain checks cover the launch-slot bundle, funding-linked wallet clusters and creator sells. RugCheck data covers insiders, holder concentration, authorities and LP lock status; DexScreener covers liquidity. Missing data is never guessed: the result is marked PARTIAL. Severe flags cap the score at 25, including a creator-sell cap: if the creator wallet's launch-slot tokens that have left that wallet (sold or moved out) exceed 20% of total supply, the score is capped at 25 (below the 75 pass line). This uses on-chain balances only and says nothing about intent.
• Jito bundle check: alongside the launch-slot bundle check, we ask the public Jito Bundle Explorer whether the token's creation transaction landed in a Jito bundle, and which launch buys shared that bundle. It is shown as “Jito bundle: yes / no / unknown” and is confirmation only: those buys are already in the launch-slot count, so it never changes the score. “Unknown” means the explorer did not answer; the launch-slot check still applies. Many tokens are created through bundles for ordinary reasons, so a yes is context, not an accusation.
• Was I the exit?: for a wallet's recent meme-coin buys, shows who sold within ±90 seconds and how the token has done since.
• Top Rugs, Dumper Wallets and Money Flow Map: a log of rugs flagged by @canarytape, wallets that sold during each crash, and where the SOL went next (up to 2 hops). Regenerated on a schedule; each table shows when it was last refreshed.
• Track record: how coins moved after we scored them. See the method below and the results.
• Venue Ratings: Solana launchpads and terminals rated on a published rubric with cited sources.

Methodology and data sources

• Solana RPC (transactions, token accounts, supply, holders), through a read-only proxy that keeps the API key server-side and does not log the addresses you look up.
• pump.fun public trade data and candles; RugCheck token reports; DexScreener pairs, paid orders and boosts; GeckoTerminal prices, trades and trending pools; public Telegram group previews (member counts only).
• Scoring runs in your browser. Every check links to a public explorer so it can be verified.

Track record method

• Every data build logs each coin's Canary Score (same 10 checks and caps as the checker), how many checks had data, its DexScreener price and market cap, and the time. The log only grows; old snapshots are never rewritten.
• For the first time we scored a coin, we take the real price from GeckoTerminal hourly candles at that time, at +24 hours and at +7 days, and report the % change. If no trade happened in an hour, the last traded price stands. Nothing is estimated: a horizon that hasn't passed yet is shown as pending, and a coin without price data is shown as such.
• Results are grouped into score ranges 0–25, 26–50, 51–75 and 76–100 with the coin count, median % change and the share that fell more than 80% (“died”). Under 10 coins is flagged as a very small sample, under 30 as small.
• Full scores have been logged since Oct 18, 2026. Earlier builds only saved some checks, so we rebuilt partial scores from them (at least 3 of 10 checks). Those are shown separately and lean generous, because missing checks don't take points off.
• So far the coins come from our rug log, so they were scored after a crash had already been reported. The track record shows what prices did next; it doesn't yet show whether a score warned anyone in advance. It is market data, not advice.

Limits

• Scores are automated heuristics. A pass does not mean a token is safe; a fail is not an accusation.
• Free data sources can be slow, rate-limited or wrong. When a check has no data, the score says PARTIAL.
• Top Rugs data is regenerated several times a day. If it is more than 24 hours old, a notice appears at the top of the page; the live checker is not affected.
• Wallet clusters are based on shared first-funding wallets. Exchanges and busy services fund many unrelated wallets, so a funder with 1,000+ transactions is ignored when it funded only two of the traced wallets. This can miss some links, and a shared funder is a fact about funding, not proof of common control.

Neutrality and corrections

We describe recorded blockchain activity, and attribute any third-party claim to its source. We do not call anyone a scammer. If you are listed and think something is wrong, contact ; we correct errors and publish responses alongside the entry.

Who runs this

CanaryTape (canarytape.tech) is an independent project, built and run by one person. It is not affiliated with any other project with a similar name.

Official accounts: X @canarytape. CanaryTape never asks you to connect a wallet, sign a transaction, or share a seed phrase or private key. The only inputs on this site are public token mint addresses and public wallet addresses. Anyone asking for more is not us.

Disclosure

The team behind CanaryTape launched $UNRUG; it is not required to use any tool and gets no special treatment in any score.