The purpose of this paper is to introduce a multi-agent model for plan synthesis in which the production of a global shared plan is based on a promising unified framework based on HTN and POP approaches. In order to take into account agents' partial knowledge and heterogeneous skills, we propose to consider the global multi-agent planning process as a POP planning procedure where agents exchange proposals and counterproposal. Each agent's proposal is produced by a relaxed HTN approach that defines partial plans in accordance with the plan space search planning, i.e., plan steps can contain open goals and threats. Agents interactions define a joint investigation that enable them to progressively prune threats, solve open goals and elaborate solutions step by step. This distributed search is sound and complete.