Contract-address (CA) detection from tweets — with DexScreener

Quick answer: Contract-address detection means reading the token address out of a post automatically, instead of you spotting it and copying it into a chart. X-Relay scans the text of every tracked post — including replies and quote tweets — for Solana, EVM and TON addresses, for $TICKER cashtags, and for the links people post instead of an address (pump.fun, DexScreener, Birdeye, Solscan, Etherscan and its chain variants). It then resolves the token on DexScreener and attaches price, liquidity, market cap, 24h volume and a chart link to the same alert.

What an address actually looks like, per chain

Each chain writes addresses in a shape you can recognise without knowing the chain in advance, which is what makes automatic detection possible:

All six EVM networks share one address format, so the shape alone cannot tell Ethereum from Base. X-Relay treats the shape as a first guess and lets DexScreener settle which chain the token actually trades on. Solana's base58 alphabet deliberately omits 0, O, I and l, which keeps it from colliding with hex — but a 44-character string that could be read either way is resolved as EVM, because that is the safer reading.

Matching is anchored so an address has to stand on its own. A base58-looking run buried inside a longer word or a URL path is not treated as a contract; without that guard, half of X's shortlinks would look like Solana addresses.

The link case — the one that quietly loses people money

This is the part most trackers get wrong. In practice a lot of accounts never post a raw address. They post a link: a pump.fun coin page, a DexScreener pair, a Birdeye token, a Solscan or Etherscan page. To a tracker that only matches address-shaped strings, those posts contain no contract at all, and the alert arrives with nothing attached.

It gets worse, because X rewrites links. Every URL in a post ships as a t.co shortlink, and even the human-readable version X provides is truncated with an ellipsis — dexscreener.com/robinhood/0xbf…. The address is genuinely not in the visible text. Recovering it means expanding the link target first and reading the address out of the URL path.

X-Relay does that expansion and pulls the contract from these link forms:

• pump.fun/coin/<address> and pump.fun/token/<address>
• dexscreener.com/<chain>/<address>
• birdeye.so/token/<address>
• Solscan token and account pages
• Etherscan, BscScan, PolygonScan, Arbiscan, BaseScan and Optimistic Etherscan address and token pages

Cashtags, and knowing when a ticker is not a call

A post does not have to carry an address to be actionable. X-Relay also reads $TICKER cashtags — two to ten alphanumeric characters starting with a letter — and resolves them through DexScreener's search. There is deliberately no blocklist of "common word" tickers: if someone typed the dollar sign, they meant the ticker, and filtering would silently drop real signal.

The one case worth handling separately is negation. A post reading "he held no $BTC or $SOL, but rather $HYPE" mentions three tickers and is bullish on exactly one. Naively taking the first match gives you $BTC and misses the actual subject. X-Relay recognises negation phrasing — no, not, never, don't, avoid, sold, dumped, skip — and excludes the tickers it governs, including through and/or/comma chains, so the token that survives is the one the author is actually talking about.

What DexScreener adds to the alert

An address on its own is not a decision. Once the token resolves, the alert carries:

• Name, symbol and confirmed chain — the definitive chain, which may correct the shape-based guess from eth to Base or Arbitrum.
• Price in USD, and 24-hour change.
• Liquidity and market cap — the two numbers that tell you whether an entry is even possible at size.
• 24-hour volume.
• Pair age — how long the pair has existed. A pair minutes old and a pair two years old deserve very different reactions to the same chart.
• Other chains — when a ticker exists on more than one chain, the alert notes where else it trades and at what market cap, so you do not buy the wrong one.
• A chart link, so the position is one tap away rather than a copy-paste.

Replies and quotes are included, and that is the point

Detection runs on every activity a tracker is watching, not just top-level posts. That matters because contracts are frequently dropped in a reply under the announcement, or in a quote of someone else's post, rather than in the announcement itself. A tracker that only forwards main posts will hand you the hype and not the address. See Reply Radar for how reply coverage works, or the tools compared if you are still choosing.

Honest limitations

Things this does not do, so you are not surprised at the wrong moment:

• Brand-new tokens DexScreener has not indexed yet resolve as an address with no market data. You still get the contract; the numbers arrive once a pair exists.
• TON raw-format addresses (the 0: hex form) are not matched. The user-friendly EQ/UQ/kQ/0Q form is what people post on X, so that is what is supported.
• An address in an image — a screenshot of a contract — is not read. There is no OCR.
• A ticker that exists on several chains is resolved to one primary pair with the others listed. It is a hint, not a verdict; check before you buy.
• Detection is not endorsement. A contract in a tweet can be a scam, a honeypot or an impostor of a real token. The alert tells you what was posted, not whether it is safe.

Keeping the noise down

If you want contracts and nothing else, point contract alerts at their own channel and leave the general timeline elsewhere, so the channel you actually watch stays readable. Per-tracker routing means different accounts can land in different places — KOLs into one channel, project accounts into another.

FAQ

Best CA tweet trackers → · Reply tracking → · Guides and comparisons → · Pricing →