New high throughput methods resulted in huge data amount in biology. The main part of related information for genomic data is the peer-reviewed papers. Text is accessible in bulky quantities, but is often messy and disorganized. The only practical answer to analyze these data sets is using computational methods that are unrestricted by the quantity of data. Hence, using text mining tools and innovative knowledge mining software that supports the curators is inevitable. Software agents are one of these kinds of tools that provide new solution for text mining. This new solution has its individual advantages in bioinformatics. Their ability to being mobile, highly customize ability and interaction aspect of agents, are some of these advantages. In this paper We offer a software agents frame for data mining in bioinformatics. Evaluation of agents performance on text mining duty is one of significant component of our workflow.