I study the potential role of young massive star clusters (YMCs) and open star clusters (OCs) in assembling stellar-mass binary black holes (BBHs) which would be detectable as persistent gravitational wave (GW) sources by the forthcoming, space-based Laser Interferometer Space Antenna (LISA). The energetic dynamical interactions inside star clusters make them factories of assembling BBHs and other types of double-compact binaries that undergo general relativistic (GR) inspiraling and merger. The initial phase of such inspirals would, typically, sweep through the LISA GW band. This fabricates a unique opportunity to probe into the early inspiraling phases of merging BBHs, which would provide insights into their formation mechanisms. Here, such LISA sources are studied from a set of evolutionary models of star clusters with masses ranging over 10(4) M-circle dot-10(5) M-circle dot that represent YMCs and intermediate-aged OCs in metal-rich and metal-poor environments of the local Universe. These models arc evolved with long-term, direct, relativistic many-body computations incorporating state-of-the-art stellar evolutionary and remnant formation models. Based on models of the local Universe constructed with such model clusters, it is shown that YMCs and intermediate-aged OCs would yield several tens to hundreds of LISA BBH sources at the current cosmic epoch with GW frequencies within 10(-3) - 10(-1) Hz and signal-to-noise ratios (S/N) > 5, assuming a mission lifetime of 5 or 10 years. Such LISA BBHs would have a bimodal distribution in total mass, be generally eccentric (less than or similar to 0.7), and typically have similar component masses, although mass-asymmetric systems are possible. Intrinsically, there would be thousands of present-day, LISA-detectable BBHs from YMCs and OCs. That way, YMCs and OCs would provide a significant-in fact, the dominant-contribution to the stellar-mass BBH population detectable by LISA. A small fraction, < 5%, of these BBHs would undergo GR inspiral to make it to LIGO-Virgo GW frequency band and merge within the mission timespan; < 15% would do so within twice the timespan. LISA BBH source counts for a range of S/N, normalized with respect to the local cluster density, are provided. Drawbacks in the present approach and future improvements are discussed.