When the user asks you to implement something complex:
First check if it's complex enough:
- Affects 5+ files? → Use planning
- Multiple sub-tasks? → Use planning
- Architectural decisions? → Use planning
Use
mcp__snipara__rlm_planwith:query: The task descriptionmax_tokens: 16000 for comprehensive plansstrategy: "relevance_first" (default) or "depth_first"
Or use
mcp__snipara__rlm_decomposeto break into chunks:query: The taskmax_depth: 2 (recommended)
Then query context for each sub-task as you implement
Example workflow:
- User: "Implement OAuth integration"
- You: rlm_plan("Implement OAuth integration", max_tokens=16000)
- Result: Sub-queries with execution order
- You: Implement each sub-query with rlm_context_query
Converted and distributed by TomeVault — claim your Tome and manage your conversions.