Tumour spheroids are a controlled 3D in vitro model for studying how externally imposed nutrient-transport conditions shape tumour growth and invasion, and recent gradient-based assays using chemotactic chambers have shown diffusion-driven anisotropic invasion and gradient-induced morphological instabilities (e.g., invasion bias, fingering/fragmentation) that can be quantified from time-lapse imaging across tumour and non-tumour cell lines [4]. Building on this paradigm, we propose an in silico workflow that couples agent-based modelling (ABM) to resolve cell-scale mechanics and phenotype switching with a continuum transport model, where nutrient and growth-factor fields are solved using a finite volume method (FVM); cells are represented as mechanically interacting agents whose proliferation, quiescence, migration, and death depend on locally sensed substrate concentrations, enabling a direct mapping between gradient strength/boundary placement and emergent invasion patterns. The extracellular matrix is modelled as a porous 3D medium and diffusion–reaction equations capture cytokines and growth factors. The workflow is implemented using BioDynaMo for ABM [1] and OpenFOAM for FVM [2], coupled via the preCICE platform [3], and is demonstrated on simulations aligned with common cancer-research assays, including a scratch (wound healing) assay for migration/metastasis and 3D spheroid configurations (free suspension or collagen matrix) used for drug screening and efficacy testing [4]. We also discuss numerical accuracy, scalability, and uncertainty quantification, and outline future developments such as automated calibration and validation workflows and related tooling for ABMs [5], alongside computational performance improvements.