Spawn Planning Agents

Use specialized agents to gather context needed for creating detailed implementation plans.

Planning Research Needs

When planning implementations, you need:

1. Existing code patterns – How is similar functionality implemented?
2. Integration points – What files will need to change?
3. Test patterns – How should tests be written?
4. Documentation – What prior decisions or research exists?

Agent Selection for Planning

Find what exists:

• codebase-locator – Find files related to the feature area
• thoughts-locator – Find existing plans, research, or tickets

Understand patterns:

• codebase-pattern-finder – Find similar features to model after
• codebase-analyzer – Understand existing architecture

Extract insights:

• thoughts-analyzer – Extract key decisions from prior research

Planning Workflow

1. Context gathering (parallel):

   • Locate existing code in the feature area
   • Find similar implementations to model after
   • Check for prior research or decisions

2. Pattern analysis (parallel, after context):

   • Analyze architecture of similar features
   • Understand integration points

3. Plan creation:

   • Use findings to inform phases
   • Reference specific files and patterns
   • Include concrete examples from codebase

Example

Planning task: “Add email notifications feature”

Step 1 – Context (parallel):

Task(subagent_type="codebase-locator", prompt="Find all notification-related code")
Task(subagent_type="codebase-pattern-finder", prompt="Find examples of background job handlers and email sending")
Task(subagent_type="thoughts-locator", prompt="Find any research about notification systems")

Step 2 – Analysis (after context):

Task(subagent_type="codebase-analyzer", prompt="Analyze how the existing SMS notification system works, including job queuing and template rendering")

Step 3 – Plan with specifics: Reference the patterns found and create phases that follow existing conventions.