You will conclude your path analysis on p values, for sure. p values are what depicts that either that hypothesized path relation was 'accepted' or 'rejected'.
I have never seen such case when p value is less than 0.05 and ZERO comes under the Lower and upper CIs and t value does not exceed 1.96. However, if you have such a case then p-value is the top notch criteria to determine.
I would report the CIs. Considering the p-values, they are derived under the assumption of normality, which might be violated in your case. Depending on the CI you use, its is more robust against the violation of certain distributional assumptions.