With the development of high throughput sequencing technologies, a diverse landscape of sequencing data has emerged. Different sequencing methods produce reads incorporating synthetic structures such as adapter sequences, barcodes and unique molecular identifiers, which must be trimmed off or extracted prior to mapping. Software tools that work with these structured reads are often tailor-made for one specific task and sequencing method. New programs must be devised whenever a new task is encountered; for example, when quantifying a new kind of artefact, or demultiplexing and trimming data from a new or niche sequencing method. There is an urgent need for flexible preprocessing software that does not sacrifice speed or simplicity to support the plurality of sequencing data, enabling bioinformaticians to fluently work with FASTQ, FASTA and BAM/SAM files without writing their own tools from scratch. To address this, we introduce Matchbox, a powerful and versatile read processing tool. By providing a rich language for error-tolerant pattern-matching and manipulation of reads, Matchbox enables flexible and fast processing of reads agnostic of the sequencing methods used to generate them. Users can write their own Matchbox scripts to tackle many bioinformatic problems, and pre-made Matchbox scripts are provided for common tasks such as trimming Illumina short-read and Oxford Nanopore Technologies long-read data, as well as demultiplexing 10X or SPLiT-seq barcodes. We demonstrate that Matchbox achieves a fast speed comparable to that of existing flexible tools on common tasks, but addresses a broader range of bioinformatic needs, representing a new state-of-the-art in sequence processing.