Conversational Agents (CAs) have become increasingly popular in many settings, including households. However, despite the increasing frequency of children’s interactions with these systems, there is still little research on the ethical design of CAs, particularly for this special population. To address this gap, in this study we design, develop and evaluate a Child-Friendly CA for collaborative storytelling, implementing specific guidelines to ensure a trustworthy design for children based on key principles such as human agency, data privacy or transparency as outlined by the High-Level Expert Group on artificial intelligence (HLEG). To evaluate the trustworthiness of the Child-Friendly CA, designers and developers conduct a collaborative assessment by applying the Assessment List for Trustworthy Artificial Intelligence (ALTAI) using the Delphi methodology. Our results demonstrate that our Child-Friendly CA design improves the trustworthiness of the system and highlights the importance of designing CAs that consider the particularities of children’s interactions. Our findings contribute to the still scarce literature on trustworthy CAs and provide insights for developers striving to ensure a trustworthy experience for children.