We have analyzed the new deep {\it XMM-Newton} and {\it Chandra} observations of the energetic radio-quiet pulsar J1813$-$1246. The X-ray spectrum is non-thermal, very hard and absorbed. Based on spectral considerations, we propose that J1813 is located at a distance further than 2.5 kpc. J1813 is highly pulsed in the X-ray domain, with a light curve characterized by two sharp, asymmetrical peaks, separated by 0.5 in phase. We detected no significant X-ray spectral changes during the pulsar phase. We extended the available {\it Fermi} ephemeris to five years. We found two glitches. The $��$-ray lightcurve is characterized by two peaks, separated by 0.5 in phase, with a bridge in between and no off-pulse emission. The spectrum shows clear evolution in phase, being softer at the peaks and hardenning towards the bridge. The X-ray peaks lag the $��$-ray ones by 0.25 in phase. We found a hint of detection in the 30-500 keV band with {\it INTEGRAL} IBIS/ISGRI, that is consistent with the extrapolation of both the soft X-ray and $��$-ray emission of J1813. The peculiar X and $��$-ray phasing suggests a singular emission geometry. We discuss some possibilities within the current pulsar emission models. Finally, we develop an alternative geometrical model where the X-ray emission comes from polar cap pair cascades. 26 pages, 10 figures, 2 tables, submitted to Astrophysical Journal
