Abstract <p>The volume of molecular dynamics (MD) simulation data shared via public repositories is rapidly increasing; however, fragmentation across multiple independent repositories, each employing distinct dataset identifiers and metadata schemas, hinders the efficient exploration and reuse of these data. In this study, we present GROMACS MetaDump, a tool for automatic annotation of output files from GROMACS MD simulations producing human- and machine-readable metadata leveraging the native GROMACS metadata (<Emphasis FontCategory="NonProportional">gmx dump</Emphasis>) output. The tool takes the run input simulation file (<Emphasis FontCategory="NonProportional">.tpr</Emphasis>) as the basis for the metadata output, optionally extending it with annotations from topology and structure files (<Emphasis FontCategory="NonProportional">.top</Emphasis>, <Emphasis FontCategory="NonProportional">.gro</Emphasis>). The tool is available as a web application (<a href="https://gmd.ceitec.cz/">https://gmd.ceitec.cz/</a>), API service, and a command-line utility. By automating the metadata extraction process, GROMACS MetaDump aims to simplify and standardise the extraction of rich, structured metadata from GROMACS MD simulations, making it easier to share, discover, and reuse simulation data within the research community.</p> Scientific contribution <p>This work introduces GROMACS MetaDump, a software tool for the automatic extraction of metadata from molecular dynamics (MD) simulations performed with GROMACS. GROMACS MetaDump captures all extractable simulation parameters, such as software version, force field, water model, box geometry, temperature, etc., and returns them in a structured JSON or YAML file. As a result, GROMACS MetaDump supports the creation of unified metadata annotations of MD simulations, making datasets indexable and findable in line with the FAIR principles.</p>