Abstract. We present the N-band (8-13 µm) spectrum of the hidden compact radio super-star cluster in NGC 5253, C2, obtained with TIMMI2 on the ESO 3.6 m telescope. The spectrum is characterised by a rising continuum due to warm dust, a silicate absorption and a strong [S ] line at 10.5 µm. Weaker lines of [Ar ] at 9.0 µm and [Ne ] at 12.8 µm are also present.The continuum can be modelled by an optically thick emission from hot (T d = 253 ± 1 K) dust emission extinguished by a cold foreground dust screen and a silicate absorption feature with A sil = 0.73 ± 0.05 mag. We show how the spatial scale of the observations greatly determine the mid-IR appearance of NGC 5253 and the important implications that this has on the interpretation of line fluxes in terms of the properties (age, IMF, etc.) of the embedded cluster. We have modelled the observed line fluxes towards C2 using photoionisation models with the most recent spectral energy distributions available to describe the integrated properties of the stellar cluster. The detailed dependence of the mid-IR lines on parameters such as the cluster age, upper mass cutoff and power law index of the IMF, as well as the local abundance, the presence of internal dust and the density structure is largely discussed. Strong constraints on the geometry based on high spatial resolution observations at different wavelengths -near-IR (HST and Keck), mid-IR (TIMMI2) and radio (VLA) -allows us to restrain the ionisation parameters to values log U ≥ −0.5 dex. This constraint on U leads to two possible solutions for the age and upper mass cutoff of C2: 1) a young (<4 Myr) cluster with a "non-standard" IMF having a low upper mass cutoff M up < 50 M ; and 2) a cluster of ∼5−6 Myr with a standard high upper mass cutoff (M up ∼ 100 M ). A young age of <4 Myr would agree with the lack of supernovae signatures in C2 and in case of being confirmed, would be the first indication for a "non-standard", low upper mass cutoff of the IMF for an individual massive cluster. An older age of ∼5-6 Myr would imply that it is possible to "contain" and hide such a compact cluster for a longer time that what it is generally thought. Arguments in favour and against these two scenarios are presented.The origin of the [O ] 25.9 µm emission measured by ISO and the possible presence of an intermediate mass black hole inside C2 are also addressed.