Planning Decision Tree

Planning Decision Tree

1. Task Breakdown Strategy

Core Logic: Stub-First & E2E

CRITICALLY IMPORTANT: You must split every functional requirement into two distinct task types:

1. Phase 1: Structure & Stubs

   • Goal: Create file structure, class definitions, method signatures, and STUBS.
   • Action: return None or hardcoded values.
   • Verification: Write E2E test that passes on these stubs (asserting hardcoded values).
   • Task Tag: [STUB CREATION]

2. Phase 2: Logic Implementation

   • Goal: Replace stubs with real logic.
   • Action: Implement algorithms, DB queries, API calls.
   • Verification: Update E2E test to assert real values. Add Unit Tests.
   • Task Tag: [LOGIC IMPLEMENTATION]

Decision Rule:

• IF task involves writing code -> Split into Stub + Impl.
• IF task is configuration/setup -> Single task.

2. Decomposition Levels

• Too Complex? (e.g., "Implement Auth") -> Break into "User Model", "Registration Service", "Login Service".
• Too Simple? (e.g., "Add one field") -> Combine with related tasks if safe.
• Dependency? -> Plan dependencies (Models, DB) BEFORE dependent services.

3. Concreteness Rules

• For New Projects: Define directory structure, class names, method signatures.
• For Modifications: Specify exact file paths and methods to change.

4. Prioritization

1. Critical: Core Architecture, Blockers, Database Schemas.
2. High: Main Business Logic (Stub -> Impl).
3. Medium: Edge cases, Error handling.