Plan Mode
Break down complex tools into step-by-step plans before generating code.
What is Plan Mode?
Plan Mode allows you to describe complex tools in natural language, and PegBoard will create a detailed implementation plan before generating any code. This ensures the AI understands your requirements correctly and breaks down complex tasks into manageable steps.
When to Use Plan Mode
- • Complex multi-step tools - Operations that require multiple phases or user interactions
- • Unclear requirements - When you want to explore different implementation approaches
- • Learning - Understanding how a tool will work before generating code
- • Team review - Share plans with teammates before committing to an implementation
Using Plan Mode
Creating a Plan
- 1
Click "New Tool" and enable "Plan Mode" toggle
- 2
Describe what you want the tool to do in natural language
- 3
Click "Generate Plan" - the AI will create a step-by-step implementation outline
Reviewing the Plan
The plan will include:
- • Overview - High-level description of the tool's functionality
- • Implementation Steps - Numbered breakdown of how the tool will work
- • API References - Specific platform APIs and functions that will be used
- • User Interactions - Dialog boxes, selections, or inputs required
- • Edge Cases - How the tool handles errors or unexpected inputs
Refining the Plan
- 1
Review the generated plan and identify any missing details or corrections
- 2
Click "Refine Plan" and provide additional context or corrections
- 3
Repeat until you're satisfied with the plan
Generating from Plan
Once you approve the plan, PegBoard generates code that follows the outlined steps.
- 1
Review the final plan and click "Approve & Generate"
- 2
PegBoard generates the tool code following the approved plan
- 3
The plan is saved with the tool metadata for future reference
Example Workflow
- Show file browser dialog to select export directory
- Display panel with naming template options (e.g., "{layer_name}", "{layer_name}_{index}")
- Iterate through all layers in current scene using
bpy.data.collections - For each layer:
- Hide all other layers temporarily
- Apply naming template to generate filename
- Export visible objects using
bpy.ops.export_scene.fbx()
- Restore original layer visibility
- Show completion message with count of exported files