The purchase and maintenance of agricultural machinery can be one of the most formidable challenges for smallholder farmers in recent years. The agricultural machinery socialization services can effectively improve the quality of agricultural production through integration and redistribution of farm machinery resources. However, the traditional agricultural machinery socialization systems tend to confine the resources within their respective individual systems. The participation of tasks and workers was limited in the matching processes of other systems. Thus, a federated platform is necessary to combine the different agricultural machinery services in order to facilitate the development of agricultural modernization by sharing the resources for cross-platform matching. But, the implementation of the federated platform can cause significant security concerns. For example, there is a potential risk of sensitive information leakage, when the data is shared across different platforms, resulting in serious privacy violations. Additionally, the centralized servers responsible for the task-matching services cannot always return accurate predicts. Therefore, it is crucial to explore safe and reliable privacy matching in the federated platform. In this study, a privacy-preserving matching scheme was introduced using CP-ABSE. The farming tasks were matched to the optimal farm mechanics. The blockchain was then employed as the underlying platform to establish the federation of agricultural machinery service platforms. The blockchain ledger was selected to record the task information of each platform and the transferring data of cross-platform, in order to avoid tampered data. Furthermore, each agricultural machinery service maintained the autonomy to access and then utilize the potential resources from other cooperation platforms. CP-ABSE technology was used to ensure the confidentiality of sensitive data in both tasks and farmers. As such, an accurate matching was achieved under ciphertext conditions. The matching farming tasks and agricultural mechanics were transformed into a task access control and keyword-based search. Specifically, the matching between farming task types and mechanics' interests could be transformed to the match between encrypted keywords and ciphertext indexes by the matching scheme. The fine-grained access control between multiple data owners and users was taken as the matching between farming operation requirements and farm services available from mechanics. Simultaneously, the service code was deployed onto the smart contracts. Different agricultural machinery service platforms published the encrypted task requirements on the blockchain using upload smart contracts. Smart contracts were then matched to perform the secure task-matching services. This approach replaced the need for a centralized server, such as the single points of failure and tampering with the match data. The security analysis showed that the matching scheme was achieved in the integrity and confidentiality of sensitive data that was shared with other cooperative platforms. Additionally, the ciphertext of tasks stored on the blockchain ledger possessed indistinguishability, which could resist the selective plaintext attacks. Finally, a prototype test system was constructed using Hyperledger Fabric, in order to verify the effectiveness of the matching scheme. The average latency of the matching smart was contracted around 250 ms, indicating that the operation efficiency of the algorithm fully met the basic application requirements of the federated platform. The scheme can be expected to effectively achieve the security of sensitive data and cross-platform task matching in the federation platforms of agricultural machinery service.