In this study, a text steganography technique suitable for Arabic texts is proposed, together with its implementation. This technique hides information by inserting extension characters (Kashida) at suitable word positions. We insert extension character in a word position to hold secret bit one and leaving position empty to hold secret bit zero. The Huffman compression algorithm is used to convert the embedding message into a compressed binary form and an Arabic text steganography technique based on character extensions is used to insert the compressed binary into the determined positions of the words in the cover text. This method was compared with some existing Arabic text stganography methods and it was shown more capacity and security.