ai-factory.implement

Sequential task execution from a development plan is automated by loading project context, processing tasks one by one, and tracking states from pending to completed. The Skill saves progress automatically so you can resume implementation across sessions.

Yes, implementation tasks can be paused and resumed in a new session. Progress persistence saves the current task states automatically, allowing you to stop development and continue the exact execution plan later without losing context.

Project context loading works by reading the DESCRIPTION.md file and applying past patches during coding plan execution. This contextual awareness ensures the implementation respects existing project constraints and prior development experience.

Git worktrees are supported for task execution and offer cleanup options upon completion. You do not strictly need them to start, but they help isolate code changes and manage commit checkpoints when implementing features systematically.

The best way to track task states during automated implementation is using a managed system that marks tasks as pending, in_progress, and completed. This enforces verbose logging requirements and updates task status as code changes occur.

Limitations of automated task plans for coding implementation include the strict requirement for verbose logging and dependency on a predefined plan. Execution is sequential, so tasks must be clearly defined upfront before the Skill can process them.