hcastro · August 15, 2025 02:05
diff --git a/gistfile1.txt b/gistfile1.txt
 This rule activates when the user asks you to enter "Planner Mode" for any task requiring non-trivial code changes, architectural design, or multi-step implementation. Planner Mode is for deeply reflective, structured planning. You must thoroughly analyze the request, audit the existing codebase or system, identify all impacted areas, and uncover hidden dependencies before proceeding. This ensures proposed plans are complete, feasible, and ready for approval. This rule should always be applied before any form of direct implementation is suggested.

 <example>
 **Valid:**  
 User: “Enter planner mode to add per-user access control to the dashboard page.”  
 Agent:  
 ✅ Asks 5 clarifying questions (e.g. "How is authentication currently managed?" "Should access be role-based or ACL?")  
 ✅ Audits the routing and middleware layers to identify affected code paths  
 ✅ Returns a 3-phase plan with file references, proposed changes, and rationale  
 ✅ Requests approval before executing  
 ✅ After phase 1, provides update: "Added middleware scaffold and tested routing fallback. Next: implement role-check logic."
 </example>

 <example type="invalid">
 **Invalid:**  
 User: “Enter planner mode to refactor this API route.”  
 Agent:  
 ❌ Proceeds to refactor immediately without analyzing broader implications  
 ❌ Asks no clarifying questions  
 ❌ Doesn’t mention which files or systems are impacted  
 ❌ Fails to present a plan or ask for approval  
 </example>
	This rule activates when the user asks you to enter "Planner Mode" for any task requiring non-trivial code changes, architectural design, or multi-step implementation. Planner Mode is for deeply reflective, structured planning. You must thoroughly analyze the request, audit the existing codebase or system, identify all impacted areas, and uncover hidden dependencies before proceeding. This ensures proposed plans are complete, feasible, and ready for approval. This rule should always be applied before any form of direct implementation is suggested.

	<example>
	Valid:
	User: “Enter planner mode to add per-user access control to the dashboard page.”
	Agent:
	✅ Asks 5 clarifying questions (e.g. "How is authentication currently managed?" "Should access be role-based or ACL?")
	✅ Audits the routing and middleware layers to identify affected code paths
	✅ Returns a 3-phase plan with file references, proposed changes, and rationale
	✅ Requests approval before executing
	✅ After phase 1, provides update: "Added middleware scaffold and tested routing fallback. Next: implement role-check logic."
	</example>

	<example type="invalid">
	Invalid:
	User: “Enter planner mode to refactor this API route.”
	Agent:
	❌ Proceeds to refactor immediately without analyzing broader implications
	❌ Asks no clarifying questions
	❌ Doesn’t mention which files or systems are impacted
	❌ Fails to present a plan or ask for approval
	</example>
No results found