Abstract: Community clouds have arisen as a promising infrastructure to face the ever-growing demand for computational power needed by applications. In such a community, different cloud providers federate their resources, allowing users to run applications across multiple sites through a global workflow scheduling system. However, current workflow scheduling approaches exhibit limited scalability, as they rely mostly on centralised scheduling mechanisms, and thus cannot fully leverage possible cooperation among cloud providers. To overcome these limitations, we propose a fully decentralised workflow scheduling system which uses a chemistry-inspired model to coordinate decision making in a community cloud platform. Our system implements a distributed shared space inside which resources are associated to tasks, enabling a decentralised decomposition of workflows into tasks. They are then independently mapped onto resources. In particular, in spite of decentralisation, the system is able to select the momentarily most appropriate resource for a given task, independently of the cloud provider the resource is located on. The framework's expectations in terms of scalability are captured through simulation experiments.