Matched‐filtering (template‐matching) is an effective method for detecting clustered seismicity such as aftershocks, low‐frequency earthquakes, repeating earthquakes, and tectonic and volcanic swarms. Several groups have developed efficient codes implementing matched‐filter methods and demonstrated that earthquake catalogs can be substantially expanded using these methods. Here, we present a near‐real‐time implementation of the matched‐filter method, designed to be used in response to ongoing seismicity. Its near‐real‐time capabilities enable dense catalogs of seismicity to be constructed rapidly, providing input into real‐time seismic hazard and forecasting and thus informing the earthquake response and scientific understanding. Such rapid development of detailed earthquake catalogs has similar application in volcano monitoring, monitoring of induced seismicity, and for online construction of slow‐earthquake catalogs. Our software package, RT‐EQcorrscan, is an open‐source extension of the EQcorrscan Python package. The package can either be deployed to apply near‐real‐time matched‐filters to a specific geographic region or sequence on a continuous basis, or configured to respond to large earthquakes or high‐rate sequences by automatically starting a matched‐filter run in response to these events. The system relies on, and maintains, a constantly updated template database of waveforms and event metadata, which is then queried for the specific target region. This template database can be updated while the matched‐filter is running to enable the set of templates to expand in response to previous results. Multiple region‐specific matched‐filters can be run in parallel, allowing the system to respond to distinct trigger events.