Read the explanation
The source regular expression accepts contiguous candidate strings of thirty two through forty four characters from a Base58-style alphabet. Bars use ten pixels per character. This is syntax screening only: no address decoding, checksum, chain lookup, token legitimacy or investment safety is verified. The code scans pasted lines rather than monitoring live accounts. Each pasted line uses its first at-sign account match. Two source lines with different handles become one retained line when one exact handle is allowed. Bars use one hundred fifty pixels per candidate line. A comma-separated allowed list is lowercased; it does not fetch posts or verify account ownership. Lines without an account are unknown unless filtered away. Pasting the same candidate twice yields two detection rows even though there is only one distinct address string. Bars compare two rows and one unique string at one hundred fifty pixels per count. The first dollar token label in each line applies to each candidate on that line. Export includes all stored rows; editing text does not rescan until Scan is clicked. Reset empties source and detections.