ULTRAPLAN Three Variants
- Entity ID:
ent-20260419-ultravar0001 - Type:
concept - Scope:
shared - Status:
active - Aliases: ultraplan variants, one_shot two_pass three_subagents_with_critique, ultraplan execution modes
Description
Three ULTRAPLAN execution modes captured by Piebald-AI tracker at v2.1.88. one_shot: single Opus 4.6 pass, 10-15 minutes, default for focused tasks. two_pass: initial analysis + refinement pass, medium complexity. three_subagents_with_critique: 3 parallel research agents exploring different codebase dimensions (risk surfaces, dependency chains, existing patterns) + 1 critique agent reviewing synthesized plan for gaps; justifies 30-minute compute budget. Auto-selection based on prompt complexity means most users invoking /ultraplan get one_shot without knowing three_subagents_with_critique exists.
Key claims
- Deep ULTRAPLAN session costs $4-8
- ULTRAPLAN auto-selects variant from prompt complexity
Relations
- piebald-ai/cc-prompts --[informed_by]--> ULTRAPLAN Three Variants
- ULTRAPLAN Three Variants --[implements]--> ULTRAPLAN