Hybrid reconfigurable intelligent surfaces (HRISs), which simultaneously provide passive reflection and active controllable amplification, are key enablers for enhancing spectral efficiency and mitigating interference in full-duplex (FD) systems. In this work, we investigate an HRIS-assisted FD system under imperfect channel state information (ICSI). A linear minimum mean square error (LMMSE) estimator is employed for channel acquisition, and the resulting signal-to-interference-plus-noise ratio (SINR) is derived. To facilitate tractable performance analysis, the SINR distribution is approximated by a Gamma distribution via moment matching, enabling closed-form expressions for both the outage probability and ergodic capacity. The analysis accounts for (i) HRIS phase errors, (ii) channel estimation errors, (iii) reflected-link interference, (iv) loopback interference, and (v) thermal noise. Monte Carlo simulations corroborate the analytical results and demonstrate that: (a) HRIS deployment yields substantial gains in outage and capacity performance, and (b) even moderate ICSI errors introduce non-negligible degradation, thereby underscoring the importance of robust channel estimation.