Total: 1
A growing number of transiting exoplanets exhibit spectroscopic evidence that their leading and trailing limbs differ. Interpreting these observations is now the limiting step. The standard approach of fitting limb-resolved transit depths results in highly anti-correlated transit depths, and treating them as independent measurements can result in strongly biased ($3σ$) retrieved molecular abundances. We present an inference framework that avoids these correlations by using the temporal signature of limb asymmetry instead. This wavelength-dependent shift in mid-transit time, or `time-shift spectrum', carries the limb-asymmetry information, is measurable from existing light curves with a spherically symmetric transit model, and is uncorrelated with the spherically-averaged transit depth. Simultaneously fitting the resulting transmission and time-shift spectrum with an independent atmospheric column for each limb enables the retrieval of limb-resolved properties from two independent measurements. We demonstrate that this approach reduces the mean squared error on the retrieved molecular abundances by factors of $\sim30-50$ relative to next best of the alternative approaches currently adopted in the field, and improves retrieved abundance precision by $\sim3-6\times$. We apply the framework to WASP-94A b and present its panchromatic limb-resolved spectrum, spanning 0.8-5~$μ$m and combining NIRISS/SOSS and NIRSpec/G395H observations. The planet's morning limb remains aerosol-obscured across this entire range, while its evening limb constrains the abundances of H$_2$O, CO$_2$, CO, and H$_2$S. We measure a limb-to-limb temperature difference of $415\pm58$~K at 1~mbar. This limb-resolved analysis finds [M/H]=+0.32$^{+0.11}_{-0.11}$ and C/O=0.724$^{+0.013}_{-0.013}$, consistent with the host star.