Progressive Decomposition Pattern

Problem

Solution

Implementation steps

1. Layer 0: Receive task, generate 3–5 high-level phases. Store as ordered queue.
2. Layer 1: Dequeue next phase, decompose into 3–7 atomic steps. Do not decompose future phases yet.
3. Layer 2: Execute atomic steps sequentially. After each step, update shared state with results.
4. Layer 3: After each phase completes, re-evaluate remaining phases against actual results — revise or discard phases that are no longer needed.
5. Terminate when phase queue is empty or goal state achieved.

Examples

• –Long-horizon research tasks where early findings change later research direction
• –Code refactoring where initial analysis reveals architecture different than assumed
• –Multi-document synthesis where first few documents reframe the question

Anti-patterns

• –Full upfront decomposition for tasks with high information-dependency
• –Skipping phase re-evaluation after completing each layer 1 execution
• –Treating decomposition output as fixed plan rather than living structure