We address the challenge of pilot design in cell-free massive multiple input multiple output (CF-mMIMO) integrated sensing and communications (ISAC) systems. We propose a novel pilot allocation framework based on manifold optimization that maximizes the system sum rate by minimizing coherence among pilot sequences, while enforcing unimodularity constraints in the frequency domain to ensure pilots are suitable for both communication and sensing tasks. Simulation results demonstrate that the proposed pilot design achieves communication performance comparable to state-of-the-art (SotA) algorithms, while delivering superior sensing capabilities due to its unimodular structure. These results highlight the potential of manifold-based pilot design for practical CF-mMIMO ISAC deployment.