Context. With more than 1000 h of observation from Feb. 2016 to Oct. 2019, the Spitzer Exploration Program Red Worlds (ID: 13067, 13175 and 14223) exclusively targeted TRAPPIST-1, a nearby (12 pc) ultracool dwarf star, finding that it is orbited by seven transiting Earth-sized planets. At least three of these planets orbit within the classical habitable zone of the star, and all of them are well-suited for a detailed atmospheric characterization with the upcoming JWST. Aims. The main goals of the Spitzer Red Worlds program were (1) to explore the system for new transiting planets, (2) to intensively monitor the planets’ transits to yield the strongest possible constraints on their masses, sizes, compositions, and dynamics, and (3) to assess the infrared variability of the host star. In this paper, we present the global results of the project. Methods. We analyzed 88 new transits and combined them with 100 previously analyzed transits, for a total of 188 transits observed at 3.6 or 4.5 μm. For a comprehensive study, we analyzed all light curves both individually and globally. We also analyzed 29 occultations (secondary eclipses) of planet b and eight occultations of planet c observed at 4.5 μm to constrain the brightness temperatures of their daysides. Results. We identify several orphan transit-like structures in our Spitzer photometry, but all of them are of low significance. We do not confirm any new transiting planets. We do not detect any significant variation of the transit depths of the planets throughout the different campaigns. Comparing our individual and global analyses of the transits, we estimate for TRAPPIST-1 transit depth measurements mean noise floors of ~35 and 25 ppm in channels 1 and 2 of Spitzer/IRAC, respectively. We estimate that most of this noise floor is of instrumental origins and due to the large inter-pixel inhomogeneity of IRAC InSb arrays, and that the much better interpixel homogeneity of JWST instruments should result in noise floors as low as 10 ppm, which is low enough to enable the atmospheric characterization of the planets by transit transmission spectroscopy. Our analysis reveals a few outlier transits, but we cannot conclude whether or not they correspond to spot or faculae crossing events. We construct updated broadband transmission spectra for all seven planets which show consistent transit depths between the two Spitzer channels. Although we are limited by instrumental precision, the combined transmission spectrum of planet b to g tells us that their atmospheres seem unlikely to be CH₄-dominated. We identify and model five distinct high energy flares in the whole dataset, and discuss our results in the context of habitability. Finally, we fail to detect occultation signals of planets b and c at 4.5 μm, and can only set 3-σ upper limits on their dayside brightness temperatures (611 K for b 586 K for c).