Object completion in 3D scanned indoor scenes remains a challenging problem, as most current approaches either focus on completing entire scenes or isolated objects. Completing objects within their scene context is still an area of active research. A key limitation of existing methods is their disregard for the scene’s environmental cues—such as walls and floors—which could provide valuable information for defining the boundaries of incomplete objects. Additionally, object completion models are often trained on synthetic datasets, where objects are neatly aligned and centred, unlike real-world scanned data that is typically unaligned. This misalignment hinders the practical application of existing models, although some approaches have attempted to address this by estimating symmetry planes. State-of-the-art (SOTA) methods also face challenges in guiding object completion, often relying on a range of potential outputs with minimal user interaction. In this work, we aim to improve the completion of objects from partially scanned indoor scenes by leveraging environmental cues to better inform the boundaries of incomplete objects. Furthermore, we introduce an interactive voxel editor that allows users to guide the object completion process toward more accurate results. Our contributions are twofold: (1) a novel boundary-defining and object-alignment method that integrates with existing object completion pipelines, and (2) the development of an interactive voxel editing tool that enhances user control over the completion process. Experimental results demonstrate the effectiveness of our approach in improving object completion in complex, real-world scanned scenes.