#task-decomposition

#task-decomposition

Backward chaining teaches complex tasks by starting with the final step, reinforcing completion first, and adding earlier steps until the full process is mastered.

Write smart AI agent specs with clear high-level vision, break tasks into smaller chunks, plan in read-only mode first, then execute and iterate continuously while staying within practical context limits.