Abstract: Open source has become a key innovation driver for core digital economy industries, such as cloud computing and artificial intelligence industries. Open source can shorten product development cycle and reduce R&D costs through distributed development and collective innovation systems. Firms increasingly emphasize on the participation of open source communities as an important innovation strategy. From a governance perspective, software development in open source communities is a large-scale knowledge-intensive project, jointly achieved by geographically dispersed developers with distinct interests, expertise and technical skills. The self-organized and loose coupling characteristics of open source communities challenge traditional hierarchy-based control mechanism. As such, how to coordinate task interdependency and boost task fulfillment is crucial to the success of open source projects as well as the prosperity of open source communities. Most existing studies focus on the centered coordination strategies devised by a focal company, such as the main contributor and the founder of open source communities, while at the micro level, how a mass of individual developers interact with each other and contribute new knowledge to the community remains underexplored. In addition, since the amount of coordination data in open source communities is huge and hard to access, extant literature has predominantly utilized survey and interview data to evaluate knowledge sharing and innovation performance of open source projects, which results may be biased towards the subjectivity of research samples. To address the above gaps, this paper employs Python programming to collect the objective data recording the coordination activities among software developers and the process of task implementation in Huawei OpenHarmony project. This study measures the fulfillment of open source tasks by the integration of newly developed code and finds that high development interdependency, manifesting as the interrelatedness between different open source tasks, is detrimental to task fulfillment. Specifically, higher development interdependency indicates greater technological complexity of open source tasks and also coordinating barriers to achieve functional compatibility between different software modules, which imposes greater knowledge integration challenges for developers. In this regard, this study provides a nuanced understanding of self-autonomy and distributed innovation paradigm of open source communities. Furthermore, this study decomposes the coordination activities into type diversity and order diversity. Specifically, type diversity refers to the diversity of categories of coordination activities involved in an open source task while order diversity is defined by the variation in the sequential transitions between different activities. These two properties of coordination activities can capture the patterns through which developers interact and achieve common ground on problem diagnoses and solutions when conducting complex open source tasks. Therefore, this study proposes that to resolve the knowledge integration challenges, developers tend to increase type diversity and order diversity of open source coordination activities when confronted with high development interdependency. In turn, both type diversity and order diversity are conducive to task fulfillment. These findings shed light on how to facilitate knowledge integration and reduce collaboration costs through a broad covering of diverse coordination activities and also frequent iteration between these activities. Lastly, this paper confirms that developer interdependency weakens the positive impacts of type diversity and order diversity of open source coordination activities. It indicates that there may exist unique coordination patterns in the context of high developer interdependency. Overall, this study provides implications for improving innovation efficiency and competitiveness of open source community.