Reconfigurable manufacturing systems can be set up to produce small quantities of different products quickly and inexpensively, making them more responsive to changing requirements than traditional mass-production systems. One aspect of reconfigurability is autonomous self-configuration of manufacturing machines. For this purpose we propose a software application called a task-to-method transformation (T2MT) system. The system takes a task description document as input, uses key characteristics of the task to generate a detailed set of working instructions, and outputs a method description document that presents the instructions in a format understood by the machine that invoked the system. The basic system architecture consists of an external interface, a task description parser, a task analyzer and a knowledge base; additional modules may be plugged in as needed. Also, the task analyzer, which represents the application-specific portion of the functionality of the system, can be reprogrammed through dynamic scripting, which makes it convenient to adapt the system for different tasks and actors. This is demonstrated by using the system to implement T2MT for two spot welding scenarios: one where the work is performed by a robot, and one where it is performed by a human worker. Once the system was in place for the first scenario, it could be reused with minor modifications for the second one, indicating that the same system architecture and components are valid for different situations where T2MT is needed.