From arete
Splits multi-faceted problems into independent sub-sessions for focused brainstorming. Use when a problem has 3+ distinct dimensions like auth redesign, DB migration, and API changes.
npx claudepluginhub jesgarram/arete --plugin areteThis skill uses the workspace's default tool permissions.
Break a multi-faceted problem into focused brainstorm sessions that can each go through the full GROUND → SHIP flow independently.
Breaks down complex problems into manageable steps, generates sequenced action plans with dependencies, identifies obstacles, and tracks execution progress.
Guides Phase 0 brainstorming: explores feature goals, success criteria, scope, compares 2-3 approaches, documents in brainstorm.md without coding.
Guides 4-phase brainstorming: assess goal clarity, understand problem, explore 2-3 approaches with pros/cons/effort, apply adversarial red-team critique before planning.
Share bugs, ideas, or general feedback.
Break a multi-faceted problem into focused brainstorm sessions that can each go through the full GROUND → SHIP flow independently.
List the distinct sub-problems. Each should be independently brainstormable:
Sub-sessions identified:
1. [Problem A] — [core tension in 1 sentence]
2. [Problem B] — [core tension in 1 sentence]
3. [Problem C] — [core tension in 1 sentence]
Which sub-problems depend on decisions from others?
Dependencies:
- Problem B depends on Problem A (schema choice affects migration)
- Problem C is independent
Recommended order:
1. Problem A (others depend on it)
2. Problem C (independent — can run in parallel)
3. Problem B (blocked by A)
Ask the user which sub-problem to tackle first. Continue the current session on that one only. Other sub-problems are parked — not forgotten, just deferred.
In later sessions, reference prior session outputs from context/exports/ for decisions that carry forward. This is how sub-sessions compose into a coherent whole.
Concise. Present the decomposition as a structured list, not a wall of text. Let the user react and choose.
Don't decompose problems that are genuinely coupled. If changing one dimension necessarily changes the others, it's one problem — not three. Ask: "Can I decide A without knowing B?" If not, they stay together.