Multi-agent systems are commonly used in various researchfields such as artificial intelligence, operational research, simulation, biol-ogy, ... However, this diversity often requires that the system and agentsin it are created from scratch for each new research project. In additionto the fact that this forces the developers to code similar elements aneweach time, this can introduce non-negligible biases (e.g. an informationaccessible to every agent which shouldn’t be or a scheduler executingtwice due to a user interface design failure). To avoid this, we proposein this paper AMAK, a framework developed in Java™to facilitate thedesign and development of a multi-agent system. First, we present theparticularity of Adaptive Multi-Agent Systems. Secondly, a state of theart of the main tools and software aiming at facilitating the developmentof such systems is discussed. Then, we develop the architecture of theframework and the main features. The use of the framework is illustratedwith an application for socio-technical ambient systems. And finally, weconclude with the perspectives of this work